The bibliometric analysis developed in this research was based on the online database of the Web of Science (https://www.webofscience.com/wos/alldb/basic-search). These search expressions included applying the terms diabetic retinopathy OR DR , which seeks words on titles, and abstracts, from 2002 to 2022. A total of 27564 publications were retrieved after refinement. All analyses were performed using the "Analyzing Results" tool provided by WoS with MS Excel support (v. 2016) to perform calculations and to develop graphs (visualization of bibliometric indicators) using the following information: i) several publications, ii) several citations iii) impact factor (for journals) and iv) h-index (for journals and authors).

The DR-related targets were screened 3 types from 9 databases sources:

(1) Database based: ①MedGene (https://www.ncbi.nlm.nih.gov/), ②Gendome (https://ai.citexs.com/), ③HGNC (https://www.genenames.org/), ④OMIM (http://www.omim.org/), ⑤GeneCards (https://www.genecards.org/), ⑥PheGenI (https://www.ncbi.nlm.nih.gov/gap/phegeni) These searches expressions consisted in applying the terms diabetic retinopathy OR DR , only Homo sapiens proteins linked to DR were selected. We got 4400 individual genes total, and details are described in Supply 1. We next use the R package UpSetR to visualise these related targets. 

(2) GEO Chip based: Using the NCBI Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/), the GSE41019, GSE60436, and GSE146615 related chips were obtained using diabetic retinopathy OR DR as the search term in Homo sapiens. The chips correspond to eye tissues and cell lines. The 3 sets of chips included DNA and RNA experiments, and their annotation files are the platform GPL6884 and GLP10558. We combined the same platform information to analyze.

(3) STRING Connected based: Cytoscape (Version 3.9.1) is an open-source software used to analyze and visualize biological networks. We installed the String and String Enrichment Apps, search tools for retrieving interacting genes or proteins. According to the String disease query and String Pubmed query, using diabetic retinopathy OR DR as the keyword, all these results were imported into Cytoscape to detect the possible relationships.

Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP, https://tcmspw.com/tcmsp.php) is a database based on the framework of systems pharmacology for herbal medicine. The CloudPhar traditional Chinese medicine (TCM-LTM, https://cloud.tasly.com/#/tcm/home) is used as the discovery of the complex mechanism of action of TCM, which contains 48126 prescriptions, 9122 plants, 34967 active ingredients, and 13109 targets for user querying the symptom, prescription, herb, and ingredient of TCM. The keyword diabetic retinopathy or diabetes + retina was searched in both 2 databases. 

Wayne diagram (http://jvenn.toulouse.inra.fr/app/example.html) was used to display the intersection of DR putative TCM. 26 herbs were obtained. Next, all these herbs were retrieved in TCMSP and screened with OB≧ 30% and LD≧ 0.18. The chemical compounds of each herb were obtained. Sorting by occurrence times, we got TOP. 10. Beta-Sitosterol is the most vital compound in TCM of TCM, which is contained in 12 of 26 kinds of DR-treatment herbs. TCMSP website was used to predict potential targets of Beta-Sitosterol for treating DR. The UniProt database (https://www.uniprot.org/) was used to convert the target name into the official name.

Next step, we matched Beta-Sitosterol treatment targets with DR targets to obtain overlapping targets. 

The core targets of Beta-Sitosterol in DR treatment were imported into the Meta-scape platform, and analyze parameters of P value < 0.01, a minimum count of 4, and an enrichment factor >1.5 (the enrichment factor is the ratio between the observed counts and the counts expected by chance) were set. Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and GSEA pathway enrichment analyses on core targets were performed.

We constructed two networks: (1) A compound-target network; and (2) A target-pathway network. Protein-protein interaction (PPI) data were attained by STRING. 

Only “Homo sapiens” targets linked to DR were selected in 6 different databases mentioned above. We got 4400 genes total, and details are described in Supplement 1. Each target got a score by the number of appearance times (Table 1, Figure 2B- Figure 2C).

We next merge GSE41019 and GSE60436 which are from the same platform. Using R package ggplot2 to visualise these chips (Figure 2D- Figure 2E). Using R package limma for the visualization results of analyzing the differential genes of microarray data (Figure 2F- Figure 2G). Figure 2

Understanding of TCM on Diabetic Retinopathy

In total, 360 and 45 kinds of herbs were obtained from the TCMSP and TCM-LTM. Paeoniae Radix Alba, licorice, and other 26 herbs were included in both 2 databases (Figure 3A, Supplement 2). According to OB ≥ 30% and DL ≥ 0.18, we listed all active compounds of these 26 kinds of herbs (Table 2, Supplement 3).

Beta-sitosterol (SIT) is a kind of sterol occurred in natural plants, like oranges, beans, and so on. It has been reported to be involved in bioactivities like anti-inflammatory, antioxidant, anticancer, angiogenic, and immunomodulatory activities. It is indicated for benign prostate hyperplasia, hypercholesterolemia, and coronary heart disease.

SIT has 2612 potential targets after removing duplicates. Based on the intersection of drug and disease targets, we obtained 23 genes (Figure 3B, Figure 5A- Figure 5B, Supplement 4), meaning these genes could play a major role in SIT treatment for DR.

GSEA analysis was used first to analyze GEO database information, which was enriched to 33 pathways (Supplement 6). The top 5 main pathways included biocarta MAPK pathway, KEGG Alanine Aspartate, and Glutamate metabolism, Reactome export of viral ribonucleoproteins from the nucleus, Reactome Gene silencing by RNA and reaction HCMV early events (Figure 4A). 

The GO and KEGG pathway enrichment analysis was performed on the total related targets in DR and the above-mentioned 7 modules (Cluster GEO, Cluster GeneDoma, Cluster OMIM, Cluster PheGene, Cluster GeneCard, Cluster Pubmed, and Cluster String). Total DR-related targets enriched to Human cytomegalovirus infection, Apoptosis, NOD-like receptor signaling pathway, and other 12 pathways (Figure 4B, Supplement 7). Because there are more than 4700 genes related in DR, we next only analyzed those scores> 4. Next high correlational genes were enriched to 10 pathways, respectively, which were enriched mainly and significantly in the AGE-RAGE signaling pathway in diabetic complications (p= 3.90E-21), Non-alcoholic fatty liver disease (p= 4.45E-21), PI3K-Akt signaling pathway (p= 1.60E-20), HIF-1 signaling pathway (p= 9.88E-18), EGFR tyrosine kinase inhibitor resistance (p= 1.77E-17) (Figure 4D). 

Next, all Clusters were enriched to 11 pathways, respectively, which were enriched respectively and mainly in the PI3K-Akt signaling pathway, positive regulation of leukocyte cell-cell adhesion, and regulation of T cell activation Figure 4.

The GO and KEGG pathway enrichment analysis was performed on the SIT (Figure 4E). Human cytomegalovirus infection (p= 1.19E-41) and hepatitis B ((p= 1.13E-46) are high enrichment levels. Lots of these pathways were relevant to inflammation.

We first use the 23 genes of both DR-related and SIT-related to enrich, PI3K-Akt signaling pathway (p= 0.035), focal adhesion(p= 3.21E-06), platelet-derived growth factor binding (PDGF, p= 0.007), extracellular matrix binding (p= 0.007) are high enrichment level (Figure 4C), which means inflammatory reaction and angiogenesis might play a vital role in SIT treat DR.

In the DR-SIT network, some signaling pathways (such as the AGE-RAGE signaling pathway in diabetics, PI3K-Akt signaling pathway, and EGFR tyrosine kinase inhibitor resistance) with higher degrees are closely related to the development of DR (Figure 4G).

We use bioinformatics analysis to screen the core genes for SIT in treating DR, ERBB3, HLA-DRB1, SPP1, CXCL10, NOX4, LOX, FBLN1, LEP, CAV1, IGF2, CD28, UCHL1, COL4A1, EDN3, BIRC5, PDGFRA, LGALS1, GSN, AURKA, TPX2, SIRPA, RIPK2, ANXA2 in the cross pathway. COL4A1, ERBB3, IGF2, PDGFRA, and SPP1 related PI3K-Akt signaling pathways or EDN3, IGF2, CXCL10, LEP, and SPP1 related receptor ligand activity pathways might be the main pharmacological targets, and pathways in DR (Fig. 5C-5D). 

Figure.Predicting pathway and target gene point. 5A. DR high correlational genes in GEO database. 5B. DR-SIT cross gene in GEO database. 5C. PI3K-Akt signaling pathways related points in GEO database. 5D. receptor ligand activity pathways in GEO database.

In the past decades, DM has imposed substantial economic and social burdens globally. DM patients are at increased risk for microvascular complications, especially retinopathy and nephropathy. (YOU CAN ADD REFERENCE OR NOT, THIS PART LOOKS MORE INTRODUCTION. TOU CAN MENTION MILLION PEOPLE WORLWIDE)

The Chinese developed TCM, a prescription of their traditional medicines, based on their own experiences of its usage, and eastern ideas. TCM nowadays has been used as a resource for developing antitumor, anti-oxidation, anti-proliferation, and anti-inflammation drugs12. It has been demonstrated by landmark studies that are using TCM oral, which was a homology of medicine and food as the main raw materials, reduces the risk of development and progression of DR9,13,14. Although traditional Chinese herbs were widely used for DM, hypertension and other chronic illness for two millennia in China, their mechanisms are unclear yet.

In this work, network analysis and bioinformatics were employed to screen the core action targets of the DR treatment. This paper uses the research ideas of network pharmacology to explore the potential drug and predict the possible targets and pathways in the SIT treatment of DR.

By searching for computer retrieval, we respectively evaluated the disease-related Databases, the TCM Databases, and the Web of Science. We first found out 26 kinds of TCM like Ginkgo Semen, Mori Fructus, Zanthoxyli Pericarpium, Portulacae Herba, and Coriandri Sativi Herba are also used as edible spice plants in the Asian diet can be used to treat DR. To explore the precise mechanisms of TCM treatment DR, we next investigated the vital active phytochemicals and molecular mechanisms of these TCM.

The big data of 9 databases sources predicted that VEGF, CTGF, KDR, PGF, ANGPT, PON1, FLT, STAT3, MMP2, TEK, FN1, FIGF are highly related to DR. PI3K-Akt signaling pathway is the highest biocarta pathway. Regulation of cell-cell adhesion, response to oxidative stress, and regulation of apoptotic, and epithelial cell proliferation play vital roles in the process of oxidative stress damage and neovascularization in the DR pathological process.

SIT is a kind of bioactive phytosterol in plant cell membranes, which has a similar chemical structure to mammalian cholesterol. European Foods Safety Authority (EFSA) suggested that 1.5-2.4 g/day of SIT reduces cholesterol levels in the blood, reducing the risk of heart disease15,16. Sutherland WH et al. reported that plasma plant sterol concentrations were significantly lower in human diabetics compared to normal controls and serum insulin levels were inversely correlated with plasma plant sterol concentration in diabetics17. Jen-Wai Chai et al. reported SIT reduced glucose uptake and adipogenesis in rat adipocytes, which means SIT can be used in DM and weight management18.

The data analysis in this paper found that, in the pathogenesis of DR, multiple targets, such as RAGE, MAPK, EGR, SOD1, TGFB1, NOD1, NLRP3, JUN, and MMP2 acted together with VEGF. Inflammatory and glycosylase activity are directly interlinked with the pharmacological action of SIT and in SIT treatment DR. The AGE-RAGE signaling pathway plays a key role in the production of DM vascular complications, especially in DR. In the cellular response to TNF-β stimulus, PDGF binding plays a central role in the process of inflammation and SIT treatment DR. COL4A1, ERBB3, IGF2, PDGFRA, and SPP1 related PI3K-Akt signaling pathway or EDN3, IGF2, CXCL10, LEP, and SPP1 related receptor ligand activity pathway might be the mainly pharmacological targets and pathways in DR.