Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Jul 22:13:907396.
doi: 10.3389/fendo.2022.907396. eCollection 2022.

CD8+T Cell-Related Gene Biomarkers in Macular Edema of Diabetic Retinopathy

Affiliations

CD8+T Cell-Related Gene Biomarkers in Macular Edema of Diabetic Retinopathy

Jing Huang et al. Front Endocrinol (Lausanne). .

Abstract

Background: CD8+T lymphocytes have a strong pro-inflammatory effect in all parts of the tissue, and some studies have demonstrated that its concentration in the vitreous increased significantly, suggesting that CD8+T cells play a pivotal role in the inflammatory response of diabetic retinopathy (DR). However, the infiltration of CD8+T cells in the DR retina, especially in diabetic macular edema (DME), and its related genes are still unclear.

Methods: Download the GSE16036 dataset from the Gene Expression Omnibus (GEO) database. The ImmuCellAI program was performed to evaluate the abundance of 24 immune cells including CD8+T cells. The CD8+T cell-related genes (DECD8+TRGs) between non-proliferative diabetic retinopathy (NPDR) and DME were detected via difference analysis and correlation analysis. Enrichment analysis and protein-protein interaction (PPI) network mapping were implemented to explore the potential function of DECD8+TRGs. Lasso regression, support vector machine recursive feature elimination (SVM-RFE), CytoHubba plug-in and MCODE plug-in in Cytoscape software, and Weighted Gene Co-Expression Network Analysis (WGCNA) were performed to comprehensively analyze and obtain Hub DECD8+TRGs. Hub DECD8+TRGs expression patterns were further validated in other two DR-related independent datasets. The CD8+TRG score was defined as the genetic characterization of Hub DECD8+TRGs using the GSVA sample scoring method, which can be administered to distinguish early and advanced diabetic nephropathy (DN) as well as normal and DN. Finally, the transcription level of DECD8+TRGs in DR model mouse were verified by quantitative real-time PCR (qPCR).

Results: A total of 371 DECD8+TRGs were identified, of which 294 genes were positively correlated and only 77 genes were negatively correlated. Eight genes (IKZF1, PTPRC, ITGB2, ITGAX, TLR7, LYN, CD74, SPI1) were recognized as Hub DECD8+TRGs. DR and DN, which have strong clinical correlation, have been proved to be associated with CD8+T cell-related hub genes by multiple independent data sets. Hub DECD8+TRGs can not only distinguish PDR from normal and DN from normal, but also play a role in the early and progressive stages of the two diseases (NPDR vs DME, Early DN vs Advanced DN). The qPCR transcription level and trend of Hub DECD8+TRGs in DR mouse model was basically the same as that in human transcriptome.

Conclusion: This study not only increases our understanding of the molecular mechanism of CD8+T cells in the progression of DME, but also expands people's cognitive vision of the molecular mechanism of crosstalk of CD8+T cells in the eyes and kidneys of patients with diabetes.

Keywords: CD8+T cell; bioinformatic analysis; biomarker; diabetic macular edema; diabetic retinopathy.

PubMed Disclaimer

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

Figure 1
Figure 1
Identification of DECD8+TRGs in macula from DR patients at NPDR stage and DME stage. (A, B) The abundance of 24 kinds of immune cells in 19 samples evaluated by ImmucellAI algorithm. (C) Overall CD8+T cell proportions comparison between 9 NPDR and 10 DME samples. (D) The volcano plot of differentially expressed genes between 9 NPDR and 10 DME samples. DECD8+TRGs, differentially expressed CD8+T cell related genes, CD8+TRGs, CD8+T cell related genes.
Figure 2
Figure 2
Enrichment analysis of DECD8+TRGs in GO, KEGG and Reactome pathways. (A) Top 30 terms of DECD8+TRGs enrichment in the biological process of GO. (B) Top 30 terms of DECD8+TRGs enriched in KEGG pathway. (C) Top 30 terms of DECD8+TRGs enriched in Reactome pathway. Fold enrichment = GeneRatio/BgRatio.
Figure 3
Figure 3
The screening process of Hub DECD8+TRGs. (A) The candidate CytoHubba genes obtained by performing eleven algorithms of the CytoHubba plug-in in Cytoscape software, the genes that meet the conditions of 9 or more algorithms were recognized as candidate CytoHubba genes and were marked in the blue box in the figure. (B) The parial likelihood deviance in the jackknife rates analysis. (C) The Lasso coefficient distribution plot of DECD8+TRGs, which is used to identify the eigenvalues of the constructed diagnostic signal. Each curve corresponds to a candidate gene. (D) SVM-RFE algorithm identifies the candidate DECD8+TRGs with the lowest error rate and the highest accuracy (10 × CV error rate = 74 -0.151). (E) The Venn plot illustrates the candidate hub genes obtained after the comprehensive intersection of CytoHubba, Lasso regression and SVM-RFE. (F) Clustering the samples and checking the outliers. Both NPDR and DME samples were included in the cluster and unsupervised clustering was performed, neither NPDR nor DME sample contained outliers. (G) The coexpression gene clusters were analyzed by dynamic tree cutting method, and the modules with high similarity were merged. The hierarchical clustering dendrogram illustrated that there were three co-expression modules, which were MEturquoise, MEblue and MEbrown. (H) Clinical feature-related gene module heat map, showing the clinical phenotypic correlation of each co-expressed module gene cluster. (I) The Venn diagram shows the genes obtained by the intersection of the candidate hub genes in figure E and the turquoise module genes of WGCNA, in which the core genes are identified by the MCODE plug-in in Cytoscape software and defined as Hub DECD8+TRGs.
Figure 4
Figure 4
Protein-protein interaction network of DECD8+TRGs. The circular diameter of the node represents the Pearson correlation coefficient between a specific gene and CD8+T cell abundance. The nodes with blue outline represent Hub DECD8+TRGs, the green nodes represent positive logFC, and the red nodes represent negative logFC. FC, fold change.
Figure 5
Figure 5
Expression pattern of DECD8+TRGs in DR. (A) Unsupervised hierarchical clustering heat map of eight Hub DECD8+TRGs in GSE160306, which illustrates the difference of z-score between NPDR and DME. (B) DECD8+TRGs were validated between normal retina and PDR fibrovascular membrane in different independent datasets. (C) The difference between normal and DME of eight Hub DECD8+TRGs in GSE160306. *p < 0.05, **p < 0.01.
Figure 6
Figure 6
Diagnostic efficacy of CD8+TRG score in patients with DN. (A, B) Comparison of CD8+TRG scores between normal and DN in GSE30528 and GSE96804 datasets. (C, D) Receiver operating characteristic curve for diagnosis of DN plotted with CD8+TRG score in GSE30528 and GSE96804 datasets. (E) Unsupervised hierarchical clustering heat map of Hub DECD8+TRGs, a set of eight genes in GSE142025, showing z-score differences between early and advanced DN. AUC, the area under the curve. ** represents p < 0.01 and *** means p < 0.001.
Figure 7
Figure 7
Hub DECD8+TRGs mRNA levels in retinas of DR model mice and control mice validated by qPCR. Compared with the control group, the transcription levels of the other seven Hub DECD8+TRGs in the DR model mouse group except Itgb2 were significantly up-regulated, while there was no significant change in Itgb2. ***p < 0.001, NS p > 0.05.

Similar articles

Cited by

References

    1. Balaji R, Duraisamy R, Kumar M. Complications of Diabetes Mellitus: A Review. Drug Invention Today (2019) 12(1):98–103.
    1. Simó R, Simó-Servat O, Bogdanov P, Hernández C. Neurovascular Unit: A New Target for Treating Early Stages of Diabetic Retinopathy. Pharmaceutics (2021) 13(8):1320. doi: 10.3390/pharmaceutics13081320 - DOI - PMC - PubMed
    1. Lent-Schochet D, Lo T, Luu K-Y, Tran S, Wilson MD, Moshiri A, et al. . Natural History and Predictors of Vision Loss in Eyes With Diabetic Macular Edema and Good Initial Visual Acuity. Retina (2021) 41(10):2132–9. doi: 10.1097/IAE.0000000000003167 - DOI - PMC - PubMed
    1. Klaassen I, Van Noorden CJ, Schlingemann RO. Molecular Basis of the Inner Blood-Retinal Barrier and its Breakdown in Diabetic Macular Edema and Other Pathological Conditions. Prog Retinal Eye Res (2013) 34:19–48. doi: 10.1016/j.preteyeres.2013.02.001 - DOI - PubMed
    1. Sun H, Saeedi P, Karuranga S, Pinkepank M, Ogurtsova K, Duncan BB, et al. . IDF Diabetes Atlas: Global, Regional and Country-Level Diabetes Prevalence Estimates for 2021 and Projections for 2045. Diabetes Res Clin Practice (2022) 183:109119. doi: 10.1016/j.diabres.2021.109119 - DOI - PMC - PubMed

Publication types