https://orcid.org/0000-0003-1094-9394
Figures
Abstract
Objectives
Diabetic retinopathy (DR) is a retinal microvascular disease associated with diabetes. Ferroptosis is a new type of programmed cell death that may participate in the occurrence and development of DR. Therefore, this study aimed to identify the DR ferroptosis-related genes by bioinformatics methods.
Methods
The RNAseq data of DR and healthy control retinas were downloaded from the gene expression synthesis (GEO) database and analyzed using the R package DESeq2. The key modules were obtained using the WGCNA algorithm, and their genes were intersected with ferroptosis-related genes in the FerrDb database to obtain differentially expressed ferroptosis-related genes (DE-FRGs). Enrichment analysis was conducted to understand the function and enrichment pathways of ferroptosis genes in DR, and hub genes were identified by protein-protein interaction (PPI) analysis. The diagnostic accuracy of hub genes for DR was evaluated according to the area under the ROC curve. The TRRUST database was then used to predict the regulatory relationship between transcription factors and target genes, with the mirDIP, ENCORI, RNAnter, RNA22, miRWalk and miRDB databases used to predict the regulatory relationship between miRNAs and target genes. Finally, another data set was used to verify the hub genes.
Results
In total, 52 ferroptosis-related DEGs (43 up-regulated and 9 down-regulated) were identified using 15 DR samples and 3 control samples and were shown to be significantly enriched in the intrinsic apoptotic signaling pathway, autophagosome, iron ion binding and p53 signaling pathway. Seven hub genes of DR ferroptosis were identified through PPI network analysis, but only HMOX1 and PTGS2 were differentially expressed in another data set. The miRNAs prediction showed that hsa-miR-873-5p was the key miRNA regulating HMOX1, while hsa-miR-624-5p and hsa-miR-542-3p were the key miRNAs regulating PTGS2. Furthermore, HMOX1 and PTGS2 were regulated by 13 and 20 transcription factors, respectively.
Citation: Huang Y, Peng J, Liang Q (2023) Identification of key ferroptosis genes in diabetic retinopathy based on bioinformatics analysis. PLoS ONE 18(1): e0280548. https://doi.org/10.1371/journal.pone.0280548
Editor: Yoshiaki Tsuji, North Carolina State University, UNITED STATES
Received: August 9, 2022; Accepted: December 19, 2022; Published: January 23, 2023
Copyright: © 2023 Huang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The GSE102485 mRNA expression profile dataset was downloaded from GEO (https://www.ncbi.nlm.nih.gov/geo/).
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
Diabetic retinopathy (DR) is a retinal microvascular disease associated with diabetes, accounting for an estimated 4.8% of global blindness, and has become the main cause of acquired visual loss in middle-aged people worldwide [1]. DR is classified as non-proliferative or proliferative DR according to the development stage [2]. Oxidative stress and inflammation play an important role in DR pathophysiology [3] but the pathogenesis and mechanism of DR remain unclear.
Ferroptosis is a new mode of programmed cell death that is different from apoptosis and involves high iron-dependent lipid peroxidation [4]. The cumulation of lipid peroxides leads to the loss of selective cell membrane permeability [5]. However, glutathione peroxidase 4 (GPX4) plays an important role in protecting cell membrane from peroxidation damage that can prevent the cell membrane from being affected by oxidation [6]. When ferroptosis occurs, the antioxidant glutathione is exhausted, leading to GPX4 failure and ultimately, fatal accumulation of lipid peroxides [7].
The high glucose environment can inhibit the growth of human retinal capillary endothelial cells, and ferroptosis can increase this effect, possibly related to GPX4 ubiquitination promoted by TRIM46 [8]. The fatty acid binding protein 4 (FABP4) can inhibit lipid peroxidation and oxidative stress by regulating ferroptosis, thereby reducing retinal damage in DR [9]. In addition, non-coding RNA (ncRNA), such as circular RNA and miRNA, also participate in the ferroptosis of DR [10, 11]; thus, ferroptosis is closely related to DR initiation and progression, but the underlying mechanism of ferroptosis in DR remains unknown.
Therefore, this study identified differentially expressed genes (DEGs) in the retina of DR patients and normal retinas and then matched these DEGs to the ferroptosis dataset to acquire differentially expressed ferroptosis-related genes (DE-FRGs) to determine the mechanism of DR development. Our findings reveal the role of ferroptosis in DR, which may become a target of clinical drug therapy in the future.
2. Materials and methods
2.1. Data collection and acquisition of ferroptosis-related gene
GEO (https://www.ncbi.nlm.nih.gov/geo/) belongs to public databases. Users can download relevant data for free for research and publish relevant articles [12]. In this study, the GSE102485 mRNA expression profile dataset was downloaded from GEO. GSE102485 contains 19 specimens of diabetic retinopathy and 3 normal retinas and is based on the GPL18573 platform (Illumina NextSeq 500, Homo sapiens). The public FerrDb (http://www.zhounan.org/ferrdb) database for ferroptosis-related genes (FRGs) that promote, inhibit, or mark ferroptosis was also searched, and after the removal of duplicate genes, 232 FRGs were finally obtained for subsequent analysis.
2.2. Identification of DEGs
R software (version 4.1.3; https://www.r-project.org/) and the Bioconductor software package (http://www.bioconductor.org/)) were used to correct and analyze the original data. RNA-seq data processing and normalization were performed using DESeq2 packages. Principal component analysis (PCA) was used to verify the repeatability of the GSE102485 data. The standard of statistical significance was |log2FC| > 1 and adjusted p < 0.05. DEGs were visualized in a volcano map based on the "ggplot2" software package.
2.3. Weighted gene co-expression network analysis
Based on the scale-free topology criterion, a co-expression network of DEGs was constructed. First, all thresholds were analyzed using the WGCNA software package in R software to determine the soft threshold power. Then, a weighted co-expression network was constructed, and the classifier was clustered into multiple modules with different colored labels to determine the correlation between each module and the control. The module most related to diabetes was considered the key module for further analysis.
2.4. Differential expression analysis of ferroptosis-related genes
The differentially expressed ferroptosis-related genes (DE-FRGs) were cross genes between key module genes and FRGs. At the same time, determine whether DE-FRGs are genes of single gene symbols. We used the "Venn Diagram" package of R software to draw Venn diagram to show the number of DE-FRGs. Used the "ggplot2" package to display the expression of DE-FRGs in the heat-map.
2.5. GO and KEGG analysis of DE-FGRs
Functional annotations and pathway enrichment for GO biological processes and KEGG annotation were performed using the “Cluster analyzer” software package. The enrichment results were sorted according to the adjusted P value. The enrichment bar chart shown the first 10 results.
2.6. Construction of a PPI network of DE-FRGs and hub gene identification
The PPI network was constructed using the search tool of STRING online database (https://cn.string-db.org/) and visualized by Cytoscape software. The top 10 genes of the PPI network were determined as the hub genes, which were calculated based on the maximal clique centrality (MCC), maximum neighborhood component (MNC), degree, edge percolated component (EPC) and Closeness algorithms y utilizing the cytohubba plug-in. The expression of hub genes was visualized in a boxplot, and the area under the curve (AUC) was measured by ROC curve analysis of hub genes. The black line diagram was drawn using the “circlize” package of R software to show the hub genes correlations.
2.7. TFs-Genes–miRNAs interaction networks
Transcriptional Regulatory Relationships Unraveled by Sentence-based Text mining (TRRUST) is the largest freely available database of human transcription factors (TFs) and genes interactions [13]. We obtained the TFs regulating hub gene from TRRUST database. Then mirDIP, ENCORI, RNAnter, RNA22, miRWalk and miRDB online databases were used to construct the miRNAs-genes cooperative regulation network. The TFs-genes-miRNAs network was visualized by Cytoscape software.
2.8. Verification of hub genes
The GSE94019 database from the GEO database was used to verify the accuracy of the identified hub genes. The "limma" package of R software was used to calibrate the data set, and differences between DR and controls were compared using a Wilcoxon test (* p-value < 0.05 and ** p-value < 0.01). The box graph was drawn using R software.
3. Results
3.1. Identification of DEGs
PCA was used to detect whether biological replicates in the same treatment group were clustered together and whether samples in different treatment groups were separated from each other. The analysis results showed that the GSE102485 data set had well repeatability (Fig 1A). It was worth noting that PCA also showed that four DR samples were outliers, so they were excluded from the analysis. Subsequently, the genes of 15 DR retinal tissues and 3 normal retinal tissues were used for differential gene analysis. Used the adjusted p value < 0.05 and |log2FC| > 1 as criterion. In all, 4876 DEGs were identified, including 2979 up-regulated genes and 1897 down-regulated genes. The differential expression of these 4876 DEGs in the DR and the control groups is shown in the volcanic map (Fig 1B).
(A) PCA for GSE102485. Red, control; Green, DR. (B) Volcano of the 4876 DEGs. Red, up-regulation; Blue down-regulation.
3.2. WGCNA analysis
The WGCNA package in R software was used to further process 4876 identified DEGs. When building the sample tree, there were no abnormal samples, so no samples were removed (Fig 2A). The soft threshold power of 18 was used to establish a scale-free co-expression network (scale-free R2 > 0.8) (Fig 2B). The clusters were divided into seven modules, blue, brown, green, red, yellow, turquoise and green, with a minimum module size of ≥ 30 (Fig 2C). The correlation between each module and DR was determined (Fig 2D), showing that blue (-0.88, P < 0.0001) and turquoise (0.95, P < 0.0001) were the most negative and positive modules related to DR, respectively. These two modules were the first two modules significantly related to clinical characteristics. The blue (cor = 0.7, P = 2.6e-151) and turquoise (cor = 0.91, P < 1e-200) modules showed a significant positive correlation between MM and GS of the target genes (Fig 2E and 2F). Therefore, blue module and turquoise module were considered as key modules.
(A) Sample clustering tree to check for outliers. (B) Scale-free networks and average connectivity for the soft threshold. (C) Clusters based on the topological overlap matrix. (D) The heat map showed the correlation of module eigengenes with DR patients or controls. (E-F) Scatter plots showing the correlations between MM and GS in the blue and turquoise modules.
3.3. Identification of DE-FRGs
The data of 232 genes from the FerrDb database were crossed with the genes of key modules to determine DE-FRGs to identify. A total of 43 positively-related genes and 9 negative related genes were found. These 52 DE-FRGs are all single gene symbols, and the heat-map and Venn diagram are shown in Fig 3. These DE-FRGs were further classified as ferroptosis driver, suppressor, or marker genes (Table 1).
(A) 3703 DEGs were intersected with the ferroptosis dataset to obtain 52 DE-FRGs. Down (green) represents the number of genes in the negative correlation module, ferroptosis (yellow) represents the number of ferroptosis genes, and Up (cobalt blue) represents the number of genes in the positive correlation module. (B) The heat map showed the distribution of 52 DE-FRGs in the DR and control groups. Blue represents down-regulated genes; red represents up-regulated genes.
3.4. GO and KEGG analysis of DE-FRGs
DE-FRGs were enriched and analyzed by R software to define their underlying physiological functions. The enrichment histogram shows the top 10 results. The most significant enrichment terms were the intrinsic apoptotic signaling pathway, cellular response to external stimulus, response to oxidative stress, response to nutrient levels(biological process) (Fig 4A), secondary lysosome, the autolysosome, autophagosome, membrane raft (cellular component) (Fig 4B), iron ion binding, antioxidant activity, ubiquitin protein ligase binding, ubiquitin-like protein ligase binding (molecular function)(Fig 4C). As expected, the KEGG pathway analysis showed that these DE-FRGs were significantly enriched in ferroptosis, the p53 signaling pathway, and endocrine resistance (Fig 4D).
(A-C) The GO analysis shows BP (Biological process), CC (Cellular component) and MF (Molecular function) respectively. (D) KEGG pathway analysis.
3.5. PPI network of DE-FRGs
Protein-Protein interaction networks were created to demonstrate the interactions between DE-FRGs (Fig 5A). We get a PPI network with 42 nodes and 115 edges. Ten of the 52 genes are not related to other molecules and do not form a molecular network. There are 47 up-regulated genes and 5 down-regulated genes in PPI network. Table 2 shows the first 10 central genes obtained in by five algorithms (MCC, MNC, Degree, EPC, Closeness). The overlapping genes in the five algorithms are selected as hub genes. They are HMOX1, PTGS2, EGFR, CAV1, TLR4, MAPK8, CDKN2A. HMOX1, PTGS2, EGFR, CAV1, TLR4, and CDKN2A are up-regulated genes. MAPK8 is a down-regulated gene.
(A) PPI network of 52 DE-FRGs. Green represents downregulated genes, and red represents upregulated genes. V shape represents the hub genes; ellipse represents others. (B) ROC curves of the seven genes for the diagnosis when distinguishing DR from normal. (C)Red represents positive correlation, green represents negative correlation, and the darker the color or the thicker the line, the higher the correlation intensity.
3.6. Hub gene detection
We discuss the diagnostic ability of these seven genes in different patients, and draw the ROC curve. The results show that the AUC of HMOX1, PTGS2, EGFR, CAV1, TLR4, MAPK8 and CDKN2A were 0.911, 1.000, 0.911, 0.978, 1.000, 1.000 and 0.889 respectively when distinguishing DR patients from normal controls (Fig 5B). The results show that 7 genes as new biomarkers have high diagnostic accuracy. The chord diagram shows that there is a strong correlation between the seven genes (Fig 5C).
3.7. Further miRNAs and TFs interaction and mining
MiRNAs-genes-TFs interactions were collected using network analysis. Hub genes (HMOX1, PTGS2, EGFR, CAV1, TLR4, MAPK8, CDKN2A) were screened for miRNAs and TFs identification (Fig 6H). CAV1 is regulated by 4 miRNAs, CDKN2A by 3 miRNAs, EGFR by 2 miRNAs, HMOX1 by 1 miRNA, MAPK8 by 3 miRNAs, PTGS2 by 2 miRNAs and TLR4 by 1 miRNA (Table 3). CAV1 is regulated by 4 TFs, CDKN2A by 8 TFs, EGFR by 17 TFs, HMOX1 by 13 TFs, MAPK8 by 2 TFs, PTGS2 by 20 TFs, and TLR4 by 1 TF (Table 4). There was a common hub gene in these TFs regulatory networks, indicating that there is a high degree of interaction between TFs and hub genes.
(A-G) The number of miRNAs that CAV1, CDKN2A, EGFR, HMOX1, MAPK8, PTGS2 and TLR4 coexist in the five databases. (H) Network of miRNAs-genes-TFs interacting with hub genes. V shape represents miRNAs; Diamond represents TFs; Ellipse represents hub genes, red represents up-regulated genes, and green represents down-regulated gene.
3.8 Verification of hub genes
The “ggpubr” package of R software was used to view the data distribution, and the values of each group centered on the median are shown in Fig 7A. This shows that the data in GSE94019 are uniform and comparable, so the samples passed the quality test. The comparison of the DEGs in the DR and the control groups showed that only two genes (HMOX1 and PTGS2) were different. The expression of hub genes in each group is shown in Fig 7B.
(A) The values of each group centered on the median. (B) Expression of hub genes in each group. *p < 0.05.
4. Discussion
Diabetic retinopathy (DR) is a special microvascular complication of diabetes and an important cause of blindness in many countries [14]. In the early stage, that is, non-proliferative DR, the blood vessels in the retina are weakened, resulting in fluid leakage, macular swelling, and eventually blurred vision, whereas, in late proliferative DR, abnormal neovascularization on the surface of the retina seriously impairs vision [8]. Ii is predicted that by 2040, the number of DR patients will reach 191million globally, bringing great pressure to individuals, families, and society [15]. Therefore, elucidating the pathogenesis of DR is important to reduce the risk of blindness in diabetic patients.
Ferroptosis is a new mode of programmed cell death involving the accumulation of lipid peroxides that damage the cell membrane and is considered different from apoptosis, necrosis, and autophagy [16]. Li et al. used bioinformatics methods to identify the key genes of ferroptosis-related to Calcic aortic valve disease and proved in vitro that these hub genes are differentially expressed in Calcic aortic valve disease and normal people [17]. Previous studies have shown that ferroptosis is associated with aging retinopathy [18], light-induced retinal degeneration [19], retinitis pigmentosa [20], and glaucoma damage [21]. Therefore, we applied bioinformatics methods to explore hub genes involved in DR ferroptosis.
The present study showed that the expression of ferroptosis genes was different between the diabetic retinopathy and normal control groups. PCA indicated that the samples of DR group and control group were obviously different. There were 52 DE-FRGs that were involved in apoptosis, iron binding, autophagy, and other processes. The PPI network identified seven hub genes related to ferroptosis in DR, including HMOX1, PTGS2, EGFR, CAV1, TLR4, MAPK8, and CDKN2A. However, after verification with another data set, only HMOX1 and PTGS2 were shown to be differentially expressed between the DR and control groups. In addition, when testing the diagnostic accuracy of hub genes, the AUC of PTGS2 was equal to 1, possibly due to the small number of samples included in the study. The AUC of a subset of genes obtained using machine learning in the study by Li et al also equals 1 [22], which may indicate that an AUC of 1 is a relatively reliable diagnostic model in small-sample studies. In the future, further verification should be conducted using a larger sample size.
Heme oxygenase-1 (HMOX1)is the rate-limiting enzyme involved in the heme oxygenase reaction pathway, degrading heme into carbon monoxide (CO), ferrous (Fe2+) and biliverdin IXα [23]. The increased HMOX1 in a mouse model reduced the neuronal cell death induced by oxidative stress [24]; however, excessive activation of HMOX1 can cause glioma cell death [25]. In retinopathy, the protective or damaging effect of HMOX1 on the retina depends on the expression of HMOX1 [26]. Our study shows that the high HMOX1 expression may be related to DR, but the role of HMOX1 in DR is still unclear. Prostaglandin endoperoxide synthase (PTGS) is a rate-limiting enzyme in the arachidonic acid synthesis of prostaglandins (PGs). It is expressed exists widely in mouse, rat and human retinas, including prostaglandin endoperoxide synthase-1 and prostaglandin endoperoxide synthase-2 (PTGS1 and PTGS2) [27]. PTGS2, which can be induced by cytokines, mitogens and endotoxins, is the immediate early genetic product of inflammation [28] and may be involved in DR by regulating the anti-angiogenic factor TSP-1 and its receptor CD36 on endothelial cells [29]. Our study suggests that PTGS2 may be involved in the process of ferroptosis causing DR. Based on the close relationship between the high glucose environment and inflammation, PTGS2 and inflammation, the specific mechanism of PTGS2 participating in DR is worthy of further study.
MicroRNA (miRNA) is a small non-coding RNA involved in post-transcriptional gene regulation by degrading or inhibiting the translation of target genes [30]. TFs as modular proteins can bind to the DNA-binding domain in the promoter region of target genes to regulate transcription [31]. In our study, hsa-miR-873-5p was the key miRNA regulating HMOX1, while hsa-miR-624-5p and hsa-miR-542-3p were the key miRNAs regulating PTGS2. TFs prediction showed that HMOX1 and PTGS2 were regulated by 13 and 20 TFs respectively. It has been shown that miR-542-3p promotes the rapid degradation of PTGS2 mRNA, which partly supports our bioinformatics prediction [32]. MiR-624-5p [33–35] and miR-873-5p [36–38] have been studied in cancer, but their role in DR has not been investigated. According to the analysis of published literature, relevant TFs PPARG [39, 40], STAT1 [41], AR [42], SP1 [43], STAT3 [44], RELA [45] and EGR1 [46] are closely related to DR. The above studies reflect that our predictions are relatively reliable. Our research suggests that TFs genes miRNAs connection may play a role in DR, but a lot of research is still needed to explore its specific mechanism.
5. Conclusion
In conclusion, we found that HMOX1 and PTGS2 are hub genes involved in ferroptosis process of diabetic retinopathy. Our study is helpful to increase the understanding of researchers on the pathogenesis of diabetic retinopathy. However, how these ferroptosis-related genes play a role needs to be explored by future in vivo and in vitro experiments.
Acknowledgments
We would like to express our gratitude to all authors for their contributions to this research.
References
- 1. Hao J, Zhang H, Yu J, Chen X, Yang L. Methylene Blue Attenuates Diabetic Retinopathy by Inhibiting NLRP3 Inflammasome Activation in STZ-Induced Diabetic Rats. Ocul Immunol Inflamm. 2019;27(5):836–43. Epub 2018/04/03. pmid:29608341.
- 2. Ting DS, Cheung GC, Wong TY. Diabetic retinopathy: global prevalence, major risk factors, screening practices and public health challenges: a review. Clin Exp Ophthalmol. 2016;44(4):260–77. Epub 2015/12/31. pmid:26716602.
- 3. Cheung N, Mitchell P, Wong TY. Diabetic retinopathy. Lancet. 2010;376(9735):124–36. Epub 2010/06/29. pmid:20580421.
- 4. Qiu Y, Cao Y, Cao W, Jia Y, Lu N. The Application of Ferroptosis in Diseases. Pharmacol Res. 2020;159:104919. Epub 2020/05/29. pmid:32464324.
- 5. Stockwell BR, Friedmann Angeli JP, Bayir H, Bush AI, Conrad M, Dixon SJ, et al. Ferroptosis: A Regulated Cell Death Nexus Linking Metabolism, Redox Biology, and Disease. Cell. 2017;171(2):273–85. Epub 2017/10/07. pmid:28985560
- 6. Yang WS, SriRamaratnam R, Welsch ME, Shimada K, Skouta R, Viswanathan VS, et al. Regulation of ferroptotic cancer cell death by GPX4. Cell. 2014;156(1–2):317–31. Epub 2014/01/21. pmid:24439385
- 7. Stockwell BR, Jiang X, Gu W. Emerging Mechanisms and Disease Relevance of Ferroptosis. Trends Cell Biol. 2020;30(6):478–90. Epub 2020/05/16. pmid:32413317
- 8. Zhang J, Qiu Q, Wang H, Chen C, Luo D. TRIM46 contributes to high glucose-induced ferroptosis and cell growth inhibition in human retinal capillary endothelial cells by facilitating GPX4 ubiquitination. Exp Cell Res. 2021;407(2):112800. Epub 2021/09/07. pmid:34487731.
- 9. Fan X, Xu M, Ren Q, Fan Y, Liu B, Chen J, et al. Downregulation of fatty acid binding protein 4 alleviates lipid peroxidation and oxidative stress in diabetic retinopathy by regulating peroxisome proliferator-activated receptor gamma-mediated ferroptosis. Bioengineered. 2022;13(4):10540–51. Epub 2022/04/21. pmid:35441580
- 10. Zhou J, Sun C, Dong X, Wang H. A novel miR-338-3p/SLC1A5 axis reprograms retinal pigment epithelium to increases its resistance to high glucose-induced cell ferroptosis. J Mol Histol. 2022;53(3):561–71. Epub 2022/03/24. pmid:35320491.
- 11. Zhu Z, Duan P, Song H, Zhou R, Chen T. Downregulation of Circular RNA PSEN1 ameliorates ferroptosis of the high glucose treated retinal pigment epithelial cells via miR-200b-3p/cofilin-2 axis. Bioengineered. 2021;12(2):12555–67. Epub 2021/12/15. pmid:34903141
- 12. Clough E, Barrett T. The Gene Expression Omnibus Database. Methods Mol Biol. 2016;1418:93–110. Epub 2016/03/24. pmid:27008011
- 13. Su W, Zhao Y, Wei Y, Zhang X, Ji J, Yang S. Exploring the Pathogenesis of Psoriasis Complicated With Atherosclerosis via Microarray Data Analysis. Front Immunol. 2021;12:667690. Epub 2021/06/15. pmid:34122426
- 14. Wong TY, Cheung CM, Larsen M, Sharma S, Simo R. Diabetic retinopathy. Nat Rev Dis Primers. 2016;2:16012. Epub 2016/05/10. pmid:27159554.
- 15. Zheng Y, He M, Congdon N. The worldwide epidemic of diabetic retinopathy. Indian J Ophthalmol. 2012;60(5):428–31. Epub 2012/09/05. pmid:22944754
- 16. Wu X, Li Y, Zhang S, Zhou X. Ferroptosis as a novel therapeutic target for cardiovascular disease. Theranostics. 2021;11(7):3052–9. Epub 2021/02/05. pmid:33537073
- 17. Li XZ, Xiong ZC, Zhang SL, Hao QY, Gao M, Wang JF, et al. Potential ferroptosis key genes in calcific aortic valve disease. Front Cardiovasc Med. 2022;9:916841. Epub 2022/08/26. pmid:36003913
- 18. Zhao T, Guo X, Sun Y. Iron Accumulation and Lipid Peroxidation in the Aging Retina: Implication of Ferroptosis in Age-Related Macular Degeneration. Aging Dis. 2021;12(2):529–51. Epub 2021/04/06. pmid:33815881
- 19. Tang W, Guo J, Liu W, Ma J, Xu G. Ferrostatin-1 attenuates ferroptosis and protects the retina against light-induced retinal degeneration. Biochem Biophys Res Commun. 2021;548:27–34. Epub 2021/02/26. pmid:33631670.
- 20. Tang Z, Ju Y, Dai X, Ni N, Liu Y, Zhang D, et al. HO-1-mediated ferroptosis as a target for protection against retinal pigment epithelium degeneration. Redox Biol. 2021;43:101971. Epub 2021/04/26. pmid:33895485
- 21. Tang J, Zhuo Y, Li Y. Effects of Iron and Zinc on Mitochondria: Potential Mechanisms of Glaucomatous Injury. Front Cell Dev Biol. 2021;9:720288. Epub 2021/08/28. pmid:34447755
- 22. Li S, Chen B, Chen H, Hua Z, Shao Y, Yin H, et al. Analysis of potential genetic biomarkers and molecular mechanism of smoking-related postmenopausal osteoporosis using weighted gene co-expression network analysis and machine learning. PLoS One. 2021;16(9):e0257343. Epub 2021/09/24. pmid:34555052
- 23. Dunn LL, Midwinter RG, Ni J, Hamid HA, Parish CR, Stocker R. New insights into intracellular locations and functions of heme oxygenase-1. Antioxid Redox Signal. 2014;20(11):1723–42. Epub 2013/11/05. pmid:24180287
- 24. Chen-Roetling J, Kamalapathy P, Cao Y, Song W, Schipper HM, Regan RF. Astrocyte heme oxygenase-1 reduces mortality and improves outcome after collagenase-induced intracerebral hemorrhage. Neurobiol Dis. 2017;102:140–6. Epub 2017/03/23. pmid:28323022
- 25. Meyer N, Zielke S, Michaelis JB, Linder B, Warnsmann V, Rakel S, et al. AT 101 induces early mitochondrial dysfunction and HMOX1 (heme oxygenase 1) to trigger mitophagic cell death in glioma cells. Autophagy. 2018;14(10):1693–709. Epub 2018/06/26. pmid:29938581
- 26. Li H, Liu B, Lian L, Zhou J, Xiang S, Zhai Y, et al. High dose expression of heme oxigenase-1 induces retinal degeneration through ER stress-related DDIT3. Mol Neurodegener. 2021;16(1):16. Epub 2021/03/12. pmid:33691741
- 27. Ju WK, Neufeld AH. Cellular localization of cyclooxygenase-1 and cyclooxygenase-2 in the normal mouse, rat, and human retina. J Comp Neurol. 2002;452(4):392–9. Epub 2002/10/02. pmid:12355421.
- 28. Dubois RN, Abramson SB, Crofford L, Gupta RA, Simon LS, Van De Putte LB, et al. Cyclooxygenase in biology and disease. FASEB J. 1998;12(12):1063–73. Epub 1998/09/16. pmid:9737710.
- 29. Sennlaub F, Valamanesh F, Vazquez-Tello A, El-Asrar AM, Checchin D, Brault S, et al. Cyclooxygenase-2 in human and experimental ischemic proliferative retinopathy. Circulation. 2003;108(2):198–204. Epub 2003/06/25. pmid:12821538.
- 30. Bushati N, Cohen SM. microRNA functions. Annu Rev Cell Dev Biol. 2007;23:175–205. Epub 2007/05/18. pmid:17506695.
- 31. Zhang HM, Kuang S, Xiong X, Gao T, Liu C, Guo AY. Transcription factor and microRNA co-regulatory loops: important regulatory motifs in biological processes and diseases. Brief Bioinform. 2015;16(1):45–58. Epub 2013/12/07. pmid:24307685.
- 32. Moore AE, Young LE, Dixon DA. A common single-nucleotide polymorphism in cyclooxygenase-2 disrupts microRNA-mediated regulation. Oncogene. 2012;31(12):1592–8. Epub 2011/08/09. pmid:21822307
- 33. Luo Y, Liu W, Tang P, Jiang D, Gu C, Huang Y, et al. miR-624-5p promoted tumorigenesis and metastasis by suppressing hippo signaling through targeting PTPRB in osteosarcoma cells. J Exp Clin Cancer Res. 2019;38(1):488. Epub 2019/12/13. pmid:31829261
- 34. Alrashed MM, Alharbi H, Alshehry AS, Ahmad M, Aloahd MS. MiR-624-5p enhances NLRP3 augmented gemcitabine resistance via EMT/IL-1beta/Wnt/beta-catenin signaling pathway in ovarian cancer. J Reprod Immunol. 2022;150:103488. Epub 2022/02/07. pmid:35124344.
- 35. Yang L, Tan W, Wei Y, Xie Z, Li W, Ma X, et al. CircLIFR suppresses hepatocellular carcinoma progression by sponging miR-624-5p and inactivating the GSK-3beta/beta-catenin signaling pathway. Cell Death Dis. 2022;13(5):464. Epub 2022/05/18. pmid:35581180
- 36. Chen Q, Lin L, Xiong B, Yang W, Huang J, Shi H, et al. MiR-873-5p targets THUMPD1 to inhibit gastric cancer cell behavior and chemoresistance. J Gastrointest Oncol. 2021;12(5):2061–72. Epub 2021/11/19. pmid:34790374
- 37. Wang Z, Liu W, Wang C, Ai Z. miR-873-5p Inhibits Cell Migration and Invasion of Papillary Thyroid Cancer via Regulation of CXCL16. Onco Targets Ther. 2020;13:1037–46. Epub 2020/02/27. pmid:32099406
- 38. Li HG, Zhao LH, Lu A, Liu JB, Su ZJ, Wang XB, et al. [The mechanism of circ_0023990/miR-873-5p/ANXA2 axis regulating radiosensitivity and development of thyroid carcinoma]. Zhonghua Yi Xue Za Zhi. 2021;101(40):3329–37. Epub 2021/11/11. pmid:34758534.
- 39. Costa V, Casamassimi A, Esposito K, Villani A, Capone M, Iannella R, et al. Characterization of a novel polymorphism in PPARG regulatory region associated with type 2 diabetes and diabetic retinopathy in Italy. J Biomed Biotechnol. 2009;2009:126917. Epub 2009/01/07. pmid:19125195
- 40. Hashemian L, Sarhangi N, Afshari M, Aghaei Meybodi HR, Hasanzad M. The role of the PPARG (Pro12Ala) common genetic variant on type 2 diabetes mellitus risk. J Diabetes Metab Disord. 2021;20(2):1385–90. Epub 2021/12/14. pmid:34900790
- 41. Shin ES, Huang Q, Gurel Z, Palenski TL, Zaitoun I, Sorenson CM, et al. STAT1-mediated Bim expression promotes the apoptosis of retinal pericytes under high glucose conditions. Cell Death Dis. 2014;5(1):e986. Epub 2014/01/11. pmid:24407239
- 42. Lin S, Peng Y, Cao M, Chen R, Hu J, Pu Z, et al. Association between Aldose Reductase Gene C(-106)T Polymorphism and Diabetic Retinopathy: A Systematic Review and Meta-Analysis. Ophthalmic Res. 2020;63(3):224–33. Epub 2020/01/22. pmid:31962334.
- 43. Gong Q, Xie J, Li Y, Liu Y, Su G. Enhanced ROBO4 is mediated by up-regulation of HIF-1alpha/SP1 or reduction in miR-125b-5p/miR-146a-5p in diabetic retinopathy. J Cell Mol Med. 2019;23(7):4723–37. Epub 2019/05/17. pmid:31094072
- 44. Wang Y, Zhai WL, Yang YW. Association between NDRG2/IL-6/STAT3 signaling pathway and diabetic retinopathy in rats. Eur Rev Med Pharmacol Sci. 2020;24(7):3476–84. Epub 2020/04/25. pmid:32329820.
- 45. Tanvir Z, Nelson RF, DeCicco-Skinner K, Connaughton VP. One month of hyperglycemia alters spectral responses of the zebrafish photopic electroretinogram. Dis Model Mech. 2018;11(10). Epub 2018/08/31. pmid:30158110
- 46. Jiewei Y, Jingjing Z, Jingjing X, Guilan Z. Downregulation of circ-UBAP2 ameliorates oxidative stress and dysfunctions of human retinal microvascular endothelial cells (hRMECs) via miR-589-5p/EGR1 axis. Bioengineered. 2021;12(1):7508–18. Epub 2021/10/06. pmid:34608841