GSP reference standard (batch no. MUST-23031310, purity >97%) was purchased from Chengdu Must Bio-technology Co., Ltd.; RPMI1640 cell culture medium (batch no. PWL021-240415) was purchased from Dalian Meilun Biotechnology Co., Ltd.; fetal calf serum (batch no. GA230622) was purchased from Gibco Life Sciences in the United States; CCK-8 reagent kit (batch no. Byt-10314), SDS-PAGE precast gel (batch no. Byt-0053A), RIPA Lysis Buffer (batch no. Byt-0013C), GAPDH antibody (batch no. AF1186), HPR-labelled secondary antibody (batch no. A0208) were purchased from Shanghai Beyotime Biotechnology Co., Ltd.; VEGF antibody (batch no. sc-57496), HIF-1α antibody (batch no. sc-13515), AKT1 antibody (batch no. sc-5298) and p-AKT1 antibody (5.Ser 473) (batch no. sc-293125) were purchased from Santa Cruz Biotechnology; All primary antibodies were diluted at a ratio of 1:1500 using antibody dilution buffer, and secondary antibodies were diluted at a ratio of 1:4000.

The chemical structure data of GSP was downloaded from Pubchem database, and then SuperPred, SwissTargetPrediction and Pharmmapper databases were used to predict the potential targets of GSP. After standardizing and compiling the targets using the UniProt database, duplicates were removed to obtain the targets of GSP.

Transcriptomics data of DU (GEO ID: GSE147890) were obtained from the GEO database, which includes transcriptomic data from skin tissues of seven control mice and six diabetic mice, as reported in the previous study. The data were divided into control and diabetic groups. Subsequently, the expression profiles were annotated using GPL571 probe data, and screened for differentially expressed genes (DEGs) with p < 0.05 and FoldChange >1.5 using the Limma package. In addition, two datasets of DU (GEO ID: GSE68183 and GSE199939) were collected by the same method for supplement; the DEGs of the three datasets were pooled to remove the duplicates and then the DU-related targets were obtained. Finally, the intersection was taken by Venn analysis to obtain the potential targets of GSP in accelerating DU wound healing.

The potential targets were imported into the STRING database, specifying the species as human and setting the confidence level to 0.7. Subsequently, the PPI network information was extracted by removing isolated protein nodes and this network information was then imported into Cytoscape software for visualization. The PPI network was also analyzed by clustering using the CytoHubba plug-in, and the top 10 targets were selected as the core targets of GSP through sequencing by the MCC algorithm for the treatment of DU. Finally, the results were visualized using the ggplot2 package.

Integration of GO and KEGG enrichment analysis is a common method to analyze the functional information and pathway mechanisms of targets. To analyze the biological processes and signaling pathways associated with the potential targets of GSP and elucidate the mechanism of GSP in the treatment of DU, GO and KEGG enrichment analyses of potential targets were performed using the Cluster Profiler package with p < 0.05, with the enrichment of at least 3 targets in each pathway as the screening conditions. Finally, TOP20 KEGG pathways and TOP30 GO terms were presented using the ggplot2 package.

In order to evaluate the interaction and binding between GSP and the core target AKT1, the protein structure file of the core target (PDB ID: 4GV1) was obtained from the PDB database and pre-docking pre-processing such as dehydrogenation and hydrogenation was performed on the core target by using the PyMol 2.5.2 software. CASTp 3.0 program was used to analyze and determine the binding pockets of the core target structure that are suitable for molecular docking, followed by molecular docking using AutoDock Vina software and finally visualization using PyMol 2.5.2 software.

CETSA is a common method used to assess receptor-ligand binding and is widely used to assess drug and protein binding. L929 cell suspension was prepared using RIPA Lysis Buffer containing protease inhibitors and then divided into control and experimental groups, which were incubated with 100 μM GSP for 2 h. After centrifugation, the cells were resuspended with PBS, after which the suspension was heated at 6 different temperature points set between 35°C–60°C. Then, the supernatant was obtained after centrifugation at 17400 g for 20 min, followed by 12% SDS-PAGE. After the end of the process, the membrane was transferred to a PVDF membrane and closed with 5% defatted goat’s milk, after which, the membrane was added to the AKT1 primary antibody for incubation at 4°C overnight. Subsequently, the membrane was washed sufficiently with TBST and then added secondary antibody to incubate at room temperature for 2 h. TBST was added to ECL luminescent solution for development after the second washing of the membrane, and the results were quantified using ImageJ software and statistically analyzed and visualized using GraphPad Prism 9.0.

L929 cells were purchased from ORiCells Biotechnology (Shanghai) Co., Ltd. and cultured in complete medium supplemented with 10% fetal calf serum, 1% penicillin-streptomycin double antibody, and 89% RPMI-1640 medium; all cellular experiments were performed without mycoplasma contamination. Referring to previous studies (Huang X. et al., 2020; Li et al., 2021), high glucose medium for culturing L929 cells was prepared by adding an additional 50 mM glucose and 100 μM palmitic acid to complete medium for the preparation of DU cell models. The incubator environment was 37 °C with a humid constant temperature containing 5% CO₂, and cell passaging was performed when cells grew to 80% confluence.

The effect of GSP on the proliferation of L929 cells was assessed using CCK8 assay. Firstly, L929 cells in the logarithmic growth phase were seeded into 96-well plates at a density of 3*103 cells/well, and the cells were divided into control, model and experimental groups. Except for the control group, the cells of each group were cultured with high glucose medium, and the experimental group was treated with different concentrations of GSP. After 24 h, each well was incubated in the incubator by adding 10 μL of CCK-8 working solution for 2 h, and then the absorbance of each well at 450 nM was detected under an enzyme marker and the results were calculated.

L929 cells in the logarithmic growth phase were taken and seeded in 6-well plates at 5*10⁴ cells/well, and the grouping and drug administration were the same as in 2.8, and concentrations of GSP were set as100 μM and 200 μM (Group of GSP-L and GSP-H). After the cells had grown all over the Petri dish, a disposable lance tip was used to scratch the cells. Subsequently, the drug-containing medium with different concentrations of GSP was added after the cell debris had been rinsed repeatedly using PBS and the cells were observed for cellular healing at 0, 8, 24, and 48 h. The results were statistically analyzed using ImageJ software.

L929 cells in the logarithmic growth phase were taken and seeded in 6-well plates at 5*10⁴ cells/well, the cell grouping and administration were the same as in 2.9. Then, total protein was extracted using RIPA Lysis Buffer and protein concentration was quantified by BCA method. Protein samples were subjected to SDS-PAGE, transferred to PVDF membrane, closed by adding 5% defatted milk for 1 h, then added primary antibody and incubated at 4 °C overnight, and washed three times with TBST; they were added secondary antibody incubated at room temperature for 2 h, washed three times with TBST, and added ECL luminescent solution for development imaging. The results were normalized by GAPDH as internal reference.

All data were processed using GraphPad Prism and ImageJ software. Differences were calculated by t-test for comparisons between two groups, and comparisons between multiple groups were assessed by one-way ANOVA. The data were presented in the form of Mean ± SEM. P < 0.05 was considered statistically significant, * indicates P < 0.05; ** indicates P < 0.01; *** indicates P < 0.001; **** indicates P < 0.0001.

A total of 538 targets of GSP were collected through SuperPred, SwissTargetPrediction and Pharmmapper databases; the variance analysis of three datasets, GSE147890, GSE68183 and GSE199939, identified a total of 10,795 DU-associated targets (Figures 1A–C). Venn analysis showed a total of 215 intersecting targets between drug and disease targets (Figures 1D, E), suggesting that GSP may accelerate DU wound healing through these 215 potential targets.

PPI network analysis was performed on 215 potential targets using the STRING database and Cytoscape 3.9.1 software, and cluster analysis was performed on the PPI network using the Cytuhubba plug-in; the results showed that the core targets of GSP for DU treatment were mainly AKT1, SRC, EGFR, PIK3R1, RHOA, CDC42 and ITGB1 (Figure 2A). Calculated by Cytohubba plug-in and ranked by MCC algorithm, the results showed that AKT1 scored 4130 and ranked first, therefore, AKT1 may be the core target of GSP for DU treatment (Figure 2B).

The items obtained from the GO enrichment analysis of potential targets included 1203 biological processes (BP), 49 cellular components (CC) and 121 molecular functions (MF). Among them, the major biological processes included small molecule catabolic processes, leukocyte migration, negative regulation of responses to external stimuli and responses to nutrient levels, etc. The CC included ficolin-1-rich granules, vesicular lumen, tertiary granules and cytoplasmic vesicle lumen, etc., and the molecular functions included nuclear receptor activity, ligand-activated transcription factor activity, protein kinase-regulated activity, and kinase-regulated activity, etc. (Figure 3A). These results suggested that GSP may exert therapeutic effects on DU by modulating biological processes such as small molecule catabolic processes and leukocyte migration.

The KEGG enrichment analysis of the potential targets identified 120 signaling pathways, and the results showed that the potential targets of GSP in treating DU mainly involved the signaling pathways of HIF-1, Rap1 and VEGF, etc. (Figure 3B). Referring to previous studies, it has been reported that the HIF-1 pathway is upstream of the VEGF pathway, and plays a role in regulating the VEGF pathway (Rattner et al., 2019; Zhang et al., 2023). Therefore, combined with the results of KEGG enrichment analysis, we speculate that GSP may play a therapeutic role in DU through the HIF-1α/VEGF axis.

The molecular results showed that the binding energy of GSP to AKT1 was −8.1 kcal/mol, which indicated that GSP could bind to AKT1 to form a stable complex. The docking of GSP and AKT1 was then visualized using PyMOL, which showed that GSP and AKT1 could form hydrogen bonds at HIS-194, GLU-191, THR-312 and LYS-276 (Figure 4A). In addition, lower thermal degradation of AKT1 protein after GSP intervention was observed by CETSA experiments, and the CETSA curve was shifted to the right, which indicated that AKT1 protein was more thermally stable after GSP intervention (Figure 4B). It suggests that there could be targeted binding between GSP and AKT1. These results indicate that GSP may accelerate DU wound healing by binding to AKT1.

The effects of GSP on high glucose-induced L929 cell proliferation were examined by CCK8 assay. The results showed that cell proliferation was significantly decreased in the DU model group, and the GSP intervention promoted cell proliferation and showed a dose-dependence within a certain range; the proliferation rate of L929 cells reached the maximum when the concentration of GSP reached 160 μM, and the ED₅₀ of GSP is 93.2Μm (Figures 5A, B). In addition, the effect of GSP on high glucose-induced migration of L929 cells was also explored by wound healing assay. The results showed that GSP significantly promoted L929 cell migration, and the wound healing rate was significantly higher than the DU model group (Figures 5C, D); these results are consistent with previous studies, suggesting that GSP may be a potential therapeutic drug for DU, and can significantly accelerate the rate of DU wound healing.

We noted from the results of KEGG enrichment analysis that HIF-1 and VEGF signaling pathways may play an important role in the acceleration of DU wound healing by GSP. A large number of previous papers reported an important role of the HIF-1α/VEGF axis in the modulation of angiogenesis. Based on that, we speculated that targeted binding to AKT1 by GSP may accelerate DU wound healing through the regulation of the AKT1/HIF-1α/VEGF axis. The expression levels of AKT1, p-AKT1, VEGF and HIF-1α in L929 cells were measured by Westen Blot; the results showed that GSP significantly increase the phosphorylation level of AKT1, whereas there was no significant change in the total protein AKT1, and the expression of VEGF and HIF-1α also increased (Figure 5E). These results suggest that targeted binding to AKT1 by GSP can accelerate DU wound healing by activating the AKT1/HIF-1α/VEGF axis, further confirming the predicted results of network analysis.