Network analysis

The STRING database compiles publicly available omics data through the latest updates, covering published genomic, transcriptomic, proteomic, and literature-mined data on human protein interactions. UniProt is an internationally recognized database for protein sequences and functional annotation, containing extensive information on the primary sequences of proteins from numerous species, biological functions, modification sites, interaction relationships, and disease-related data. This study uses the UniProt and STRING databases, integrating data results through April 1, 2026 and encompassing long-term accumulated research on molecular regulation. Using SIRT1, oxidative stress, and inflammatory response as the core search keywords, functional proteins related to oxidative stress and inflammation were retrieved from UniProt, with priority given to proteins directly associated with SIRT1 and stress-induced vascular senescence phenotypes. These were combined with SIRT1 to construct an interaction network, and key targets were selected based on betweenness centrality, retaining high-confidence molecules that were directly bound or indirectly regulated while removing redundant targets with low relevance. Cytoscape was used to complete network visualization and topology analysis.

This study used the STRING V12.0 database, restricting the species to Homo sapiens and setting the interaction confidence score threshold to the database’s default high-confidence threshold of 0.7. Evidence channels included experimental validation, database annotation, gene co-expression, and literature text mining. Seven rounds of target expansion were conducted based on node-by-node association rules to comprehensively identify downstream and upstream cascade regulatory targets of SIRT1, fully covering oxidative stress- and inflammation-related signaling pathways while avoiding the omission of key molecules. The maximum number of interacting proteins per round was limited to 20, and SIRT1’s direct and indirect regulatory molecules were mined layer by layer to clarify the logic of multi-level molecular regulatory networks. The resulting protein interaction network was imported into Cytoscape software, and topological analysis was conducted using the CytoNCA plugin. In the software’s Analyze Network function, betweenness centrality was chosen as the core evaluation metric, and the Pearson’s correlation coefficient of protein interactions was used to evaluate the closeness of molecular associations. Nodes were ranked by descending correlation coefficient, with node color intensity and size representing the strength of interactions.

Based on the UniProt 2026-01 database, target proteins were identified using standard GO functional terms for oxidative stress and inflammatory response together with official keywords. The selection criteria were restricted to human-derived, reviewed (Swiss-Prot), high-quality annotated proteins, excluding low-confidence unannotated sequences and unrelated cross-species proteins. The screened proteins related to oxidative stress and inflammation were co-imported with SIRT1 into STRING to build a specific regulatory network. Node association weights and centrality parameters were recalculated in Cytoscape to finalize the hierarchical visualization analysis.

Gene Ontology (GO) enrichment analysis provides a straightforward annotation of gene products. The selected SIRT1-related genes were entered into the DAVID database, with “H. sapiens” selected as the organism, and GO analysis was carried out according to the biological process (BP), cellular component (CC), and molecular function (MF). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis emphasizes the interaction networks of molecules within the organism. Through this analysis, the signaling pathways of target molecules can be identified. Relevant information for KEGG enrichment analysis was obtained from DAVID. Based on the p-values, the top 10 data points with the smallest values were selected. After organizing the data, microinformatics was used for visualization. As shown, A denotes a BP, B denotes a CC, C denotes an MF, and D denotes KEGG pathway analysis.

Analysis of the protein interaction network

By searching the STRING database platform, we constructed a protein-protein interaction network for “SIRT1” to identify protein targets that interact with it directly or indirectly. The STRING database was queried by entering “SIRT1” and setting the species to “H. sapiens” while keeping the default parameters unchanged. After seven rounds of expansion, 80 interacting proteins were identified. The protein interaction network was then imported into Cytoscape for calculation and analysis. The protein interactions were imported into Cytoscape, and the molecules were arranged counterclockwise according to their correlation coefficients. For a redder color of the molecule, the area is larger, and the correlation with other protein molecules is stronger (Figure 3). Among them, 62 molecules (77.5 %) were directly related to SIRT1, and 18 molecules (22.5 %) were indirectly related. There were 35 molecules related to oxidative stress (44 %), primarily including “EP300,” “CREB1,” “HDAC1,” “NCOR1,” “FOXO1,” “JUN,” “NRF1,” “BCL2,” “CDKN1A,” “TP53,” “SUV39H1,” “BRCA1,” “MDM2,” “NCOR2,” “PPARG,” “PPARA,” “PPARGC1A,” “HMGB1,” “ESRRA,” and “MAPK8”. There were 27 (33.6 %) molecules related to inflammation, mainly including “EP300,” “PPARG,” “CTNNB1,” “CREB1,” “NDN,” “AKT1,” “HDAC1,” “HMGB1,” “RELA,” “FOXO1,” “MAPK8,” “MAPK1,” “JUN,” “HSP90AA1,” “HSPA9,” “BCL2,” “CREBBP,” “NCOA1,” “CDKN1A, and “FOXO3.” The network interaction graph was further analyzed using the CytoNCA tool in Cytoscape, with proteins color-coded and arranged counterclockwise according to betweenness centrality. The top 20 proteins ranked by betweenness centrality were selected for plotting, revealing that SIRT1 had the highest betweenness centrality. Among these top 20 proteins, 16 were associated with oxidative stress, and 14 were associated with inflammation (Figure 4).

Figure 3:: Global protein-protein interaction network of SIRT1.

Figure 4:: Screening of core targets in the protein interaction mapping of SIRT1.

Cytoscape analysis

Searching the UniProt platform using the term “Oxidative stress” yielded proteins associated with oxidative stress, and these data were entered into STRING to construct a protein-protein interaction network. The network was then imported into Cytoscape, where the top 20 oxidative stress-related proteins ranked by betweenness centrality were identified. These proteins, together with SIRT1, were simultaneously entered into STRING to obtain a protein interaction network map, which was imported into Cytoscape to calculate correlation coefficients and rank them counterclockwise by correlation (Figure 5A). Protein molecules directly related to SIRT1 included “BCL2,” “INS,” “PINK1,” “TP53,” “SOD2,” “SOD1,” “GPX6,” “GPX2,” “GPX3,” “GPX5,” “GPX8,” “PXDN,” and “JUN,” accounting for 65 %. Molecules indirectly related to SIRT1 included “TXNRD1,” “UBB,” “UBC,” “RPS23,” “PARK7,” “FYN,” and “POMC,” accounting for 35 %. We searched the UniProt platform and entered “inflammation” to identify inflammation-related proteins and then entered the related proteins into STRING to obtain the protein interaction network. This network was imported into Cytoscape, and the top 20 proteins associated with inflammation were identified and ranked by betweenness centrality. These proteins, together with SIRT1, were entered into STRING to generate an interaction network, which was imported into Cytoscape for correlation coefficient calculations and arranged counterclockwise according to relatedness (Figure 5B). SIRT1 was directly associated with molecules such as “IL4,” “IL18,” “AGT,” “VCAM1,” “IL1B,” “IL10,” “TLR4,” and “STAT1,” accounting for 40 %, and indirectly associated with “ARG2,” “PLA2G3,” “CDH5,” “GATA3,” “CTSG,” “ARG1,” “CTH,” “TNFAIP3,” “HCK,” “PLA2G10,” “HP,” and “AIM2,” among others, accounting for 60 %.

Figure 5:: Protein-protein interaction map of SIRT1 (A involved in oxidative stress; B involved in inflammation).

Figure description: Part A is interaction map of SIRT1 with oxidative stress related proteins, contains TP53, PXDN, SOD, GPX2, INS, etc.; part B is interaction map of SIRT1 with inflammation related proteins, contains IL1B, TLR4, IL-10, STAT1, IL-4, VCAM1, etc. Node size and color gradient stand for interaction weight. e.g. TP53, PDXN, SOD2 and SOD1 are the molecules with higher interaction weight involved in oxidative stress and SIRT1. IL1B, TLR4, IL-10, STAT1, and IL-4 are the molecules with higher interaction weight involved in inflammation and SIRT1.

Gene ontology and KEGG pathway enrichment analyses

The GO analysis results showed that the numbers of target genes interacting with SIRT1 classified into BP, MF, and CC were 191, 90, and 33, respectively. These target genes were mainly distributed in the plasma membrane, cytoplasm, and nucleus; they were primarily involved in protein binding, DNA binding, and RNA polymerase II regulation; and they participated in BPs such as signal transduction, positive regulation of transcription, and the inflammatory response. KEGG pathway enrichment analysis revealed that SIRT1 is closely associated with stress-induced aging signaling pathways (Figure 6).

Figure 6:: SIRT1 enrichment bubble chart (A shows BP involvement; B shows MF involvement; C shows CC involvement; and D shows KEGG involvement).

Figure description: Four bubble subplots, A to C compose the data for the Gene Ontology enrichment analysis, and D is for Kyoto Encyclopedia of Genes and Genomes enrichment analysis. A shows the biological process features representative terms including signal transduction, inflammatory response, etc. B is for molecular function containing protein binding, DNA binding, etc. C shows the cellular component covering cytoplasm, nucleus and other cell compartments. D presents the stress-induced aging signaling pathway and other related KEGG entries. Bubble area corresponds to gene number; color gradient visualizes −log₁₀(P value).