Introduction

Central retinal artery occlusion (CRAO) is an ophthalmic emergency characterized by acute, painless, monocular vision loss1. Thrombotic or embolic occlusion of the retinal artery can irreversibly damage the retina within 90–240 min2. Patients with CRAO have an increased risk for future cardiovascular events, such as stroke and myocardial infarction, as well as increased all-cause mortality3. Given that spontaneous recanalization of the central retinal artery rarely occurs, rapid treatment is critical4. However, current treatments demonstrate suboptimal efficacy. Even with timely treatment, further eye tissue damage persists under reperfusion, leading to unsatisfactory visual prognoses5. Further research into the pathogenesis of CRAO is warranted to explore more effective therapies and improve patient outcomes.

In addition to vascular factors, other related molecular mechanisms contribute to CRAO progression and outcomes, including key pathways such as the AKT1, JAK-STAT, MAPK, and HIF1A pathways6,7,8. Retinal ischaemia‒reperfusion leads to early stabilization of hypoxia-inducible factors and upregulation of hypoxia-responsive genes9. Transient increases in CCL2 and decreases in IL-27 occur early on. This is followed by late activation of astrocytes and Müller cells, as well as progressive inner retinal thinning. Molecular analysis revealed early decreased expression of the retinal ganglion cell marker Thy1 and late decreased expression of the amacrine cell marker Sncg, indicating ischaemic damage and loss of these neuronal cells over time9. Recent studies have also demonstrated that transcriptome analysis of peripheral blood monocytes can reveal differences in immune status between individuals and can assist in the diagnosis and prediction of eye diseases10,11. However, research into the underlying mechanisms of CRAO is limited by the scarcity of human retinal samples. Many studies have relied on animal models or circulating blood, but transcriptomic approaches to identify CRAO-related genes in human samples are lacking.

To address this gap in knowledge, this study aimed to elucidate CRAO mechanisms using RNA-seq transcriptomic profiling of accessible samples. Through transcriptome technology, key transcription factors and signalling pathways that play crucial roles in the differentiation of retinal cells have been identified12. Additionally, transcriptomic analysis can identify hub genes and key pathways with potential clinical utility13,14,15,16 and provide targets for disease treatment17. Blood samples from CRAO patients and control individuals were subjected to RNA sequencing and bioinformatic analysis to reveal differentially expressed genes (DEGs) and dysregulated pathways, which may contribute to retinal damage in patients with CRAO. Achieving these aims will provide insights into the pathogenesis of CRAO and establish a foundation for developing targeted therapies.

Material and methods

Patients recruitment and sample collection

We collected blood samples from 16 CRAO patients and 9 cataract controls at Wuhan People's Hospital in China. CRAO patients were included if they had clinical and angiographic diagnosis of CRAO less than 72 h after symptom onset, best-corrected visual acuity between counting fingers and 20/50, and availability of peripheral vein or arterial embolus site samples before thrombolysis. Patients were excluded if they had glaucoma, retinal or disc neovascularization, any previous CRAO treatment, non-CRAO vascular retinopathy, intraocular surgery in the prior 3 months, stroke, myocardial infarction, hemi-CRAO, macula-sparing CRAO, or other severe ocular diseases. The CRAO cohort aligned with published diagnostic guidelines18 including sudden vision loss, positive RAPD, retinal ischemic edema, and delayed arterial filling on angiography. Cataract controls had surgery during the same period. Additional exclusions for both groups were prior retinal artery occlusion, coronary artery disease, malignancy, severe renal insufficiency (eGFR < 30 mL/min), liver disease, stroke, or lung disease. All arterial CRAO blood samples were collected during central retinal artery interventional surgery. The ethics committee of Wuhan People's Hospital approved this study (protocol code, WDRY2022-K278 and approval date, 30 November 2022), which adhered to the 1995 Declaration of Helsinki guidelines. All participants provided informed consent.

Peripheral blood mononuclear cells (PBMCs) isolation and RNA sequencing library construction

Fresh venous anticoagulant blood samples (VBS) or arterial anticoagulant blood samples (ABS) for CRAO were obtained and blood mononuclear cells were isolated using a Ficoll separation gradient. Total RNA was extracted using Trizol reagent. RNA quality control was performed to assess purity, concentration, integrity and quantity. For RNA sequencing, 3 μg of total RNA was used as input. mRNA was purified and fragmented. cDNA was synthesized and amplified to generate sequencing libraries. Libraries were quantified and sequenced on the DNBSEQ-T7 platform using 150 bp paired-end reads. The serum samples used in this study were collected by EDTA-coated procoagulant tubes and centrifugated before frozen at − 80 °C.

Following final library quality control on the Agilent Bioanalyzer to confirm the expected size distributions, libraries were pooled and sequenced on the Illumina HiSeq3000 platform in a 150-bp paired-end read run format. Raw RNA sequencing (RNA-seq) reads were preprocessed and determined using featureCounts function in SubReads package v1.5.3 with default parameter (Parameter setting is -t exon -g gene id -s 0 -p > {log} 2 > &1).

Bioinformatic analysis of microarray result

Differential expression analysis and identification of RNAs

Quantile normalization and subsequent data processing were performed using the 'limma' package in R software (version 4.2.3)19. Principal Component Analysis (PCA) was conducted to investigate group differences based on the variables of interest. For 3D PCA analysis and plot generation, the 'stats' and 'ggplot2' packages in R were utilized. Volcano plots were employed to identify significant differentially expressed genes (DEGs) between the Artery group and the Vein group, as well as between the Vein group and the control group. Genes with |log2 fold change|≥ 2 and adjusted p-value < 0.05 were considered statistically significant in both the A–V and V–C groups. Additionally, hierarchical clustering was performed to reveal distinguishable RNA expression patterns among the samples, and the 'pheatmap' package was used to visualize the expression patterns of the 50 most significantly differentially expressed genes (DGEs) in a heatmap.

GO and KEGG enrichment analysis

The Gene Ontology (GO; http://geneontology.org) is a structured and computable knowledge resource that aids in identifying the molecular function (MF), biological process (BP), and cellular component (CC) attributes of genes20. It is particularly useful for analyzing high-throughput genome or transcriptome data. The Kyoto Encyclopedia of Genes and Genomes (KEGG; https://www.kegg.jp/) is a meticulously curated resource that enables systematic analysis of gene functions and associations by linking gene sets to corresponding pathways21. GO annotation and KEGG pathway enrichment analyses were performed using the 'clusterProfiler' package in R (version 4.2.3)22. The enrichment functions and pathways of the DEGs were determined by applying a threshold of adjusted p < 0.05 and a count of enriched genes greater than five.

Gene set enrichment analysis

The study employed the Gene Set Enrichment Analysis (GSEA) program (version 4.3.2) with default settings to investigate significant functional and pathway differences between the Artery groups and Vein groups, as well as between the Vein groups and the control group. The objective was to elucidate the underlying molecular distinctions23. Pathways enriched for each phenotype were obtained from the h subset (h.all.v2023.1.Hs.symbols.gmt) set, which served as the preset gene sets for enrichment analysis. Enriched pathways were classified based on criteria such as the adjusted p-value (< 0.05), FDR q-value (< 0.25), and normalized enrichment score (|NES|> 1).

Building protein–protein interaction networks and identifying hub genes

A protein–protein interaction (PPI) network was constructed to investigate the interaction among DEGs. The Search Tool for the Retrieval of Interacting Genes (STRING) database was utilized to provide a comprehensive score for each PPI relationship pair24. Statistically significant interactions were identified by applying a combined score threshold of 0.7. The PPI pairs with an interaction score > 0.7 were extracted and visualized using Cytoscape 3.9.025. To explore the most three key functional modules, the Cytoscape plug-in molecular complex detection (MCODE) was applied to the PPI network. Furthermore, KEGG and GO analyses of the identified modular genes were performed in R and the results were visualized using bar plots.

Selection and analysis of hub genes

The CytoHubba plug-in of Cytoscape identified hub genes from the PPI network, and then five algorithms (Radiality, MCC, MNC, Degree and EPC) were used to confirm the final hub genes, which were illustrated in Venn diagram. Ultimately, the above hub genes were submitted to GeneMANIA (http://genemania.org) to construct a co-expression network.

Immune cell infiltration analysis

Immune cell infiltration was assessed using the Cibersort algorithm in R software26. The mRNA expression matrix (normalized) was used as input to calculate the cell abundance score for 22 cell types. Stacked bar plots were generated by Chiplot (https://www.chiplot.online). Box plots were generated using the 'ggplot2' package, while heatmaps were created using the 'pheatmap' package.

Statistical analysis

SPSS (version 27.0; IBM, New York, United States) was used to perform statistical analysis. Continuous data were expressed as mean ± SD and compared by using ANOVA. Categorical variables were presented as counts (frequency) and compared using chi-square test or the Fisher exact test, as appropriate. In some analyses, more than 20% of the cells had expected frequencies less than 5, precluding calculation of p-values. A two-sided p < 0.05 was considered to statistical significant.

Ethical approval

The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethical Committee Board of Renmin Hospital of Wuhan University (protocol code, WDRY2022-K278 and approval date, 30 November 2022).

Informed consent

As this was a retrospective observational study, patient consent was waived.

Results

Clinical characteristics of the recruited patients

Table 1 presents the baseline characteristics of the study participants. A total of 25 patients were included: 16 RAO patients (9 peripheral and 7 arterial) and 9 cataract control individuals. The RAO patients and control individuals were well matched in terms of age (average 63 years), sex (64.00% male), and blood pressure (p > 0.05 across groups).

Table 1 Comparison baseline demographic data of patients (the values are presented as mean ± SD or as percentages).
Full size table

DEGs between retinal arterial obstruction patients and control individuals

RNA-seq analyses of ABS, VBS, and control venous blood were conducted to evaluate gene expression changes in PBMCs. Principal component analysis (PCA) revealed a clear separation between CRAO arterial samples and venous samples (A–V group), as well as between CRAO venous samples and control samples (V–C group) (Fig. 1a and b). In the A–V group, 1400 genes were upregulated and 1291 genes were downregulated compared to those in the control group (Fig. 1c and e). In the V–C group, 112 genes were upregulated and 34 genes were downregulated compared to those in the control group (Fig. 1d and f). Functional enrichment analysis provided insights into the biological functions associated with these DEGs.

Figure 1
figure 1

The DEGs between CRAO patients and controls. (a) PCA plot of arterial vs venous blood samples. Red dots represent CRAO arterial blood samples. Blue dots represent CRAO venous blood samples. (b) PCA plot of venous vs control blood samples. Blue dots represent CRAO venous blood samples. Yellow dots represent cataract control blood samples. (c) Volcano plot showing the DEGs of A–V group. (d) Volcano plot showing the DEGs of V–C group. (e) Heat map of DEGs in A–V group. (f) Heat map of DEGs in V–C group.

Full size image

Functional enrichment analysis of DEGs

GO and KEGG analyses were performed on the DEGs to determine their biological functions and investigate their roles. DEGs were hierarchically clustered based on GO categories: biological process (BP), molecular function (MF), and cellular component (CC). The top 10 enriched GO and KEGG pathways were identified for each category (Fig. 2a–d). Hierarchical clustering of GO categories revealed enriched pathways in both the A–V and V–C groups, including “ribosomes”, “cytoplasmic translation”, “chromatin remodelling”, “NAD(P)H dehydrogenase (quinone)”, “antigen processing and presentation” and other processes. Thus, immune system involvement, leukocyte activation, and potential immune dysfunction pathways were enriched in CRAO blood samples. KEGG analysis of both datasets also revealed transcriptional regulatory pathways, including “transcriptional misregulation in cancer”, “viral life cycle”, and “type I diabetes mellitus”, as well as “VEGF signalling pathway”, "yersinia infection" and other immune/inflammatory pathways. The V–C group exhibited enrichment of additional immune/inflammatory pathways, such as “Toll-like receptor signalling” and “Th17 cell differentiation”. GSEA of DEGs revealed enrichment of pathways involved in chromatin segregation, including CENP-A-related pathways and “chromosome, centromeric core domain”. Furthermore, GSEA revealed enrichment of multiple MHC-related genes and antigen-presenting activity pathways in the V–C group, indicating that inflammatory responses play a role in CRAO (Fig. 2e–g). In summary, both groups demonstrated ribosomal pathway enrichment, and the V–C group uniquely displayed enrichment in antigen presentation and inflammatory pathways.

Figure 2
figure 2

Functional enrichment of DEGs. GO functional analysis of DEGs in A–V group (a) and V–C group (b). KEGG functional analysis of DEGs in A–V group (c) and V–C group (d). The gene sets enriched in comparison to ABS (e) and VBS (f) in the GSEA results. (g) GSEA results indicate activation of the antigen-presenting activity pathway in the V–C group (INTERFERON_ALPHA_RESPONSE: NES = 2.56 p-value < 0.001 FDR q-value < 0.001; INTERFERON_GAMMA_RESPONSE: NES = 2.23 p-value < 0.001 FDR q-value < 0.001).

Full size image

Protein‒protein interaction (PPI) network analysis and hub gene filtration

To investigate PPIs among proteins encoded by the DEGs in CRAO, we constructed a PPI network using STRING (interaction score > 0.7) and visualized it in Cytoscape (the top 2,000 DEGs ranked by |log2-fold change| were analysed in the A–V group due to the large number of genes). Significant gene clustering modules were identified using the MCODE plug-ins (Fig. 3a and b). Enrichment analysis of the A–V group modules revealed their involvement in antigenic peptide synthesis, transcriptional regulation, protein synthesis, and NADH dehydrogenase activity (Fig. 3c). The V–C group exhibited a similar enrichment pattern, with additional emphasis on antigen recognition/presentation and MHC-associated responses (Fig. 3d). These findings indicate that ribosome-related genes and inflammatory responses may significantly contribute to CRAO development and progression.

Figure 3
figure 3

The PPI network. (a) The PPI network of common DEGs, and top three gene clustering modules in MCODE analysis of A–V group. (b) The PPI network of common DEGs, and top three gene clustering modules in MCODE analysis of V–C group. (c and d) GO and KEGG enrichment analysis of total modular genes.

Full size image

Selection and analysis of hub genes

To identify hub genes in CRAO, we used the CytoHubba plug-in and diverse algorithms to determine hub genes from the common DEGs (Tables 2 and 3). The notable hub genes of the A–V group were RPS2, RPS5, RPS9, and RPS14 (Fig. 4a). The hub genes of the V–C group were HLA-F, RPS25, RPS18, HLA-DMA, PSMB9, RPL36A, HLA-B, RPS7, RPL7A, HLA-E, TAPBP, RPS27, HLA-DMB, and RPL17 (Fig. 4b). We then analysed the coexpression network and related functions of these hub genes based on the GeneMANIA database, which revealed the top 20 most interacting genes. In the A–V group, the hub genes primarily functioned in various protein synthesis processes, including cytosolic ribosomes, ribosomes, protein targeting to the ER, and rRNA processing (Fig. 4c). The V–C group hub genes exhibited diverse functions, including immune response (T-cell-mediated immunity, antigen processing/presentation via MHC class Ib) and regulatory roles in MHC protein complexes, cytosolic ribosomes, ribosomes, and gene transcription/protein synthesis. These genes actively regulated cellular processes linked to gene expression, protein synthesis, and the immune response (Fig. 4d).

Table 2 The top 20 hub genes rank of A–V group in CytoHubba.
Full size table
Table 3 The top 20 hub genes rank of V–C group in CytoHubba.
Full size table
Figure 4
figure 4

(a) The Venn diagram identified 3 candidates for hub genes in A–V group by five algorithms. (b) The Venn diagram identified 3 candidates for hub genes in V–C group by five algorithms. (c and d) The co-expression genes were analyzed via GeneMANIA. The predicted genes are located in the outer circle, and hub gene is in the inner circle.

Full size image

Immune cell infiltration analysis

To investigate the associations between CRAO and immune cell populations in peripheral blood, the proportions of 22 immune cell types in the ABS, VBS, and CON groups were quantified using CIBERSORT (Fig. 5a–d and Supplementary Figure S1). Compared to those in the VBS group, the numbers of neutrophils and CD4+ T memory cells were markedly increased in the ABS group (Fig. 5a and c). In addition, compared to that in the CON group, the number of naive CD4+ T cells was markedly increased in the VBS group (Fig. 5b and d). The GSEA results and gene information related to the antigen presentation pathway are shown in Supplementary Figure S2, Supplementary Table S1 and Supplementary Table S2.

Figure 5
figure 5

Composition of infiltrating immune cells in CRAO. (a) The proportion of immune cell populations in A–V group was determined by CIBERSORT. (b) The proportion of immune cell populations in V–C group was determined by CIBERSORT. (c) The proportion of immune cell populations in A–V group was determined by CIBERSORT. Red parts represent CRAO arterial blood samples. Blue parts represent CRAO venous blood samples. (d) The proportion of immune cell populations in V–C group was determined by CIBERSORT. Red parts represent cataract control blood samples. Blue parts represent CRAO venous blood samples.

Full size image

Discussion

The study of genes related to CRAO is crucial, yet limited research has hindered progress. Expanding research efforts using transcriptomic profiling is vital for identifying new biomarkers and therapeutic targets. For example, a recent RNA-seq study provided the first insights into retinal transcriptional changes in animal models of acute retinal artery ischaemia27. It identified potential therapeutic targets like CXCL8 inhibition to mitigate retinal damage27. This study demonstrated the ability of transcriptomics to elucidate molecular mechanisms and develop treatment options for CRAO. Further transcriptomic research will enable better characterization of the complex CRAO gene expression landscape and improve diagnosis and management.

In this study, we identified 2691 DEGs in the A–V group and 146 DEGs in the V–C group by analysing two sets of genetic data. GO enrichment analysis revealed that the DEGs in both groups were enriched mainly in ribosomes and cytoplasmic ribosomes. The A–V group DEGs were associated with immune system activation, and the V–C group DEGs were associated with antigen presentation and binding. KEGG results suggested that these genes were primarily enriched in ribosomes and the Toll-like receptor signalling pathway. Interestingly, both the GO and KEGG pathway analyses suggested that ribosome-related genes may be involved in the pathogenesis of CRAO.

Hub gene and coexpression network analyses revealed multiple ribosomal proteins, including RPS2, RPS5, RPS7, RPS9, RPS14, RPS18, RPS27, RPL7A, RPL17, and RPL36A, as hub proteins. These proteins regulate the cell cycle, differentiation, apoptosis, and development28,29,30,31. RPS732, RPS933, and RPS2734,35,36 specifically participate in the p53 and NF-κB pathways, which are important for cellular stress and inflammation. Ischaemic conditions upregulate various ribosomal proteins in the central nervous system, such as RPS1, RPS3, RPS6, and RPS9. RPS337,38 and RPS639 demonstrate neuroprotective effects. This ribosomal protein upregulation provides insights into the enrichment of ribosomal pathways in CRAO. Similarly, in myocardial infarction, the death of resident macrophages induces neutrophil and monocyte-derived macrophage repopulation of the infarct area40, causing inflammation, cytokine release, and post-MI cell death41. However, as tissue repair is initiated, macrophages switch from proinflammatory to pro-healing phenotypes, promoting collagen deposition, matrix remodelling, and scar formation42,43. This macrophage phenotype change may explain the observed ribosomal protein enrichment in CRAO.

The upregulation of ribosomal proteins observed in CRAO likely represents an endogenous protective response, as ribosomal proteins play important cytoprotective roles in other ischaemic conditions. Ribosomal protein S6 phosphorylation impacts neuronal survival and axon regeneration in cerebral ischaemia‒reperfusion injury44. As metal binding proteins, the ribosomal proteins S6, S3a, and S4X regulate ribosomal RNA stability and are linked to cellular oxidative stress and inflammation45. Additionally, exogenous ribosomal proteins can alleviate ischaemia/hypoxia-induced cell damage. A peptide from ribosomal L23a inhibits myocardial apoptosis and ischaemia‒reperfusion injury46. Ribosomal L26 binds p53 mRNA to promote p53 translation in hypoxia-induced apoptosis46. In summary, ribosomal proteins mediate cell stress responses, growth regulation, and cell survival in ischaemic diseases. Thus, modulating their function may mitigate ischaemia‒reperfusion injury.

The major histocompatibility complex (MHC) comprises highly polymorphic gene clusters encoding proteins involved in intercellular recognition and antigen presentation and is called the HLA complex in humans. Direct allorecognition involves recipient T cells recognizing alloantigens presented by endothelial cell HLA molecules47. While most studies have focused on immune rejection post transplantation, little research has been conducted on ischaemic diseases48,49. Recent evidence indicates that in patients with ischaemic stroke, CD8+ and CD4+ T cells are activated in the cerebrospinal fluid, and T-cell migration and infiltration are guided by HLA molecule interactions50. PSMB9, located in the MHC class II region, regulates apoptosis and MHC class I antigen presentation by encoding the immunoproteasome catalytic subunit LMP2. LMP2 attenuates ischaemia/hypoxia-induced blood‒brain barrier injury through Wnt/β-catenin signalling activation51. Transporter associated with antigen processing (TAP) binding protein (TAPBP) encodes tapasin, a key MHC-I pathway component52,53, that mediates inflammatory injury via CD8+ T-cell infiltration52,54. A novel partial MHC II structure, DRmQ, promotes ischaemic stroke neuroprotection by inhibiting neuroantigen-specific T cells and blocking cytotoxic monocytes/macrophages55. DRmQ can also reduce inflammation and infarct volume in ischaemia by inhibiting T-cell activation and cytokine secretion56. These findings underscore the roles of MHC-related genes such as PSMB9 and TAPBP in regulating the early inflammatory response of CRAO through MHC-mediated T-cell activation.

Our studies revealed an imbalance between neutrophils and CD4+ T cells in patients with CRAO, suggesting that inflammatory responses and immune cell infiltration may contribute to CRAO progression57,58. Research has shown that CD4+ T cells are activated postmyocardial infarction to promote myocardial wound healing59. Additionally, CD4+ T cells can delay the ischaemic infiltration of B cells and alleviate poststroke cognitive decline60, potentially playing an important neuroprotective role. In summary, our findings imply that immune activation participates in the pathogenesis of CRAO. CD4+ T cells may promote inflammatory responses in the early stages of CRAO but also mediate repair and protection in later stages.

The present study, however, has several limitations. First, due to limitations in collecting clinical samples, retinal tissue samples from patients with CRAO were not available, and PBMCs were used as a substitute. Therefore, future studies should aim to obtain patient cooperation and consent. Second, although key pathways, hub molecules, and immune cells identified in this study were identified in human specimens, additional laboratory studies, particularly animal experiments, are necessary to confirm their pathogenic significance. Furthermore, large-scale clinical trials are needed for further investigation of the underlying mechanisms involved. Finally, this study could not determine whether the apoptosis pathway induces immune cell infiltration or confirm the involvement of immune cells in the apoptotic process. Further research is warranted to elucidate the potential mechanisms involved.

Conclusions

This transcriptomics study identified ribosomal proteins, HLA genes, and antigen processing molecules as potential biomarkers for CRAO screening and prognosis. Their dysregulation can collectively result in retinal cell dysfunction, immune-mediated damage, and CRAO progression. Specifically, decreased ribosomal protein synthesis may impact ischaemic retinal injury through insufficient production of proteins required to maintain retinal cell function61,62. Aberrant MHC gene expression can also dysregulate antigen presentation, activate proinflammatory T cells50,63, and exacerbate ischaemia‒reperfusion damage. Therefore, normalizing the expression of ribosomal proteins and MHC genes may mitigate retinal damage in patients with CRAO by attenuating cytotoxic inflammation. In summary, these transcriptomic signatures provide clinically relevant insights into CRAO mechanisms and offer biomarker-guided avenues for personalized diagnostic and therapeutic strategies. Validating the functional significance and clinical utility of these biomarkers will be key to translating these findings into precision medicine approaches for improved CRAO management.