Network pharmacology study of the mechanism underlying the therapeutic effect of Zhujing pill and its main component oleanolic acid against diabetic retinopathy

Abstract Diabetic retinopathy (DR) is the leading cause of blindness in the working population worldwide, with few effective drugs available for its treatment in the early stages. The Zhujing pill (ZJP) is well-established to enhance the early symptoms of DR, but the mechanism underlying its therapeutic effect remains unclear. In the present study, we used systems biology and multidirectional pharmacology to screen the main active ingredients of ZJP and retrieved DrugBank and Genecards databases to obtain ‘drug-disease’ common targets. Using bioinformatics analysis, we obtained the core targets, and potential mechanisms of action of ZJP and its main components for the treatment of DR. Molecular docking was used to predict the binding sites and the binding affinity of the main active ingredients to the core targets. The predicted mechanism was verified in animal experiments. We found that the main active ingredient of ZJP was oleanolic acid, and 63 common ‘drug-disease’ targets were identified. Topological analysis and cluster analysis based on the protein–protein interaction network of the Metascape database screened the core targets as PRKCA, etc. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis showed that these core targets were significantly enriched in the pro-angiogenic pathway of the VEGF signaling pathway. Molecular docking and surface plasmon resonance revealed that ZJP and its main active component, oleanolic acid had the highest binding affinity with PKC-α, the core target of the VEGF signaling pathway. Animal experiments validated that ZJP and oleanolic acid could improve DR.


Introduction
Diabetic retinopathy (DR) refers to retinal damage caused by chronic progressive diabetes mellitus, which has an insidious onset. Current evidence suggests that DR patients are largely in the middle to late stages when detected clinically. DR is widely recognized in ophthalmology as a microvascular complication of diabetes mellitus [1]. However, in terms of the entire retinal tissue structure, neuronal-glial cells account for approximately 95-97% (the remainder are neurovascular units), and their pathological damage is important during the pathogenesis of DR [2,3]. Hence, in DR, microangiopathy coexists with neuronal lesions, and treating microvascular damage alone is often insufficient [4]. It is well known that the main feature of diabetes is hyperglycemia (disorders of glucose metabolism) and that the risk of DR progressively increases with the duration of diabetes, yet strict glycemic control does not alter this phenomenon [5]. When the duration of diabetes exceeds 10 years, roughly 50% of patients have DR as a complication [6]. According to the International Diabetes Federation (IDF) 2021 data, about 537 million adults (20-79 years old) are OD = OD 620-740 nm ). Moreover, the concentration of Evans blue in the supernatant was calculated by standard curve analysis.

Histopathological observation of the retina
Paraffin-embedded tissue sections with a thickness of 5 μm were taken and stained with hematoxylin-eosin (HE) to observe the retinal tissue structure using a light microscope; another tissue section was treated with Masson staining to observe the inner boundary membrane tissue morphology.

Immunohistochemistry of VEGF in retinal tissue
Paraffin-embedded tissue sections with a thickness of 5 μm were taken for routine dewaxing, antigen repair and staining (the sections must be kept in a wet state at any time during the production process). After sealing the sections, observation was performed under a microscope and results were statistically analyzed.

Western blot was used to detect pathway and protein expression changes
The total protein was extracted from each group of rat retinal tissues, quantified by the BCA method, and 30 μg per group was taken by SDS-PAGE gel electrophoresis, transferred to PVDF membrane after electrophoresis was completed, and closed in skimmed milk for 1 h. The tissues were incubated in primary antibody diluted at 1:1000 overnight, washed 3 times with TBST, and incubated in secondary antibody diluted at 1:10 000 for 1 h at room temperature, and HIF-1α, VEGFA, PKC-α, and ERK1/2 protein expression were detected in each group, and β-Tubulin was used as the internal reference.

SPR analysis of the binding ability of drug small molecules and proteins
Protein microarray fixation When the temperature of the computer and SPR system to be tested reached the preset temperature of 25 • C, the SPR control software was started and connected to the host system. Then, according to the isoelectric point of PKC-α protein, the optimal pH value and concentration range of protein fixed on the chip were screened to ensure that more protein was fixed on the protein chip.

Oleanolic acid test solution preparation
Oleanolic acid was modified with polyethylene glycol (M r = 5000) to improve its solubility and availability at preset temperatures. PEG-OA was accurately weighed and then prepared into 20 nM small molecule stock solution with PBS solution, diluted into 10, 5, 2.5 and 1.25 nM reference solution with mobile phase solution.

Affinity detection
The LMW kinetics/affinity multi-cycle module of SPR system software was used; the binding and dissociation times were set to 60 and 180 s, respectively. After dissolution correction, the response signals recorded during protein binding and dissociation from the small-molecule drug were fitted to the data results using Biacore S200 Evaluation Software. SPSS 20.0 statistical software was used for data analysis, and data were expressed as 'mean + − standard deviation' (x + − s). The differences between groups were analyzed by one-way ANOVA, and a P-value < 0.05 was statistically significant.

Screening of putative bioactive compounds and targets of ZJP
A detailed composition of ZJP is shown in Table 1. The active ingredients of ZJP were screened in the TCMSP database using the parameters oral-bioavailability (OB) ≥ 30%, drug-likeness (DL) ≥ 0.18. A total of 42 main active ingredients were obtained.
The SDF structures of 42 active ingredients of ZJP were obtained using the PubChem database, and 116 drug targets with scores >0.1 were obtained from the SwissTarget Prediction database. Using Cytoscape 8.0.0 software, a 'drug-target' network diagram was constructed, and the Network Analyzer function was used to analyze the main active ingredients of ZJP ( Figure 2A). The circles represent the active ingredients of ZJP (n=42), and the squares are the targets (n=192). The area and color of the graph deepen and become larger with increasing degree value,  indicating the more important the ingredient. ZJP1 (Oleanolic acid), ZJP42 (Sophranol), ZJP4 (Stigmasterol), and ZJP16 (β-carotene) had the highest degree values, especially oleanolic acid (OA) ( Figure 2B).

Screening of common drug targets of active components of ZJP and diseases
The keywords 'diabetic retinopathy' were searched in OMIM, Disgenet, and Genecards databases, and a target score ≥ median value was used to identify significant disease targets. After combining and de-weighting the obtained disease targets, 1150 disease targets were screened. The intersection of the screened active ingredient targets of ZJP and its disease targets during DR treatment yielded 63 common targets, visualized in a Venn plot generated by Microscience software.

Construction of the protein-protein interaction network
The 63 common targets were imported into the Metascape platform, and 'homo sapiens' was selected to analyze the common targets. We first determined all statistical enrichment terms (GO terms and KEGG pathways), calculated the cumulative hypergeometric p-value and enrichment factor, and used them for filtering. Subsequently, according to the kappa statistical similarity between gene members, the remaining essential terms were clustered into a tree.Kappa was screened with a threshold of 0.3 and plotted. Subsequently, we selected a subset of representative terms from this cluster and converted them into a network layout (Supplementary Figure S1A and B). Specifically, each term is represented by a circular node whose size is proportional to the number of input genes in the term, and its color indicates its cluster identity (i.e., nodes of the same color belong to the same cluster). Terms with a similarity score > 0.3 were connected by an edge (the width of the edge represents the similarity score). The network was visualized using Cytoscape, which adopts a 'force-oriented' layout with clear edge binding.

Core target screening based on topology analysis
The PPI network retrieved from the Metascape platform was then imported into Cytoscape 8.0.0 for optimization (Supplementary Figure S1C). The topology analysis was performed by the Network Analyzer tool, and the four parameters, degree, betweenness centrality, average shortest path length and closeness centrality, were used as reference standards to obtain genes with scores greater than the average score as core targets, including MAPK3, RELA, ESR1, PRKCA, BCL2, HIF-1α, PTGS2, IL6, VEGFA, and so on.

Core target screening based on cluster analysis
The constructed PPI network was imported into Cytoscape 8.0.0, and the analysis of gene clusters and the screening of core targets were executed using the MCODE module. A total of three gene clusters and three core genes were obtained, and the core genes were CTNNB1, CCND1, and HIF-1α (Supplementary Figure S1D).

GO annotation and KEGG pathway enrichment analysis
GO analysis of the 63 common targets was conducted using R language and divided into biological processes, cellular components and molecular functions ( Figure 3A). GO annotation showed significant enrichment in a total of  Hydrogen bonding Hydrophobic interaction π-π interaction Binding energy (kcal mol -1 ) A total of 125 KEGG pathways were significantly enriched, and the top 10 results ( Figure 3B) included pathways in cancer (hsa05200), prostate cancer (hsa05215), small cell lung cancer (hsa05222), colorectal cancer (hsa05210), aldosterone-regulated sodium reabsorption (hsa04960), VEGF signaling pathway (hsa04370), focal adhesion (hsa04510), thyroid cancer (hsa05216), toxoplasmosis (hsa05145), and PPAR signaling pathway (hsa03320). Based on our PPI network and GO annotation results, the VEGF signaling pathway was regarded as a key mechanism for ZJP and its main active ingredient, oleanolic acid, in treating DR ( Figure 3C).

Molecular docking verification
Molecular docking of PKC-α, ERK1/2, VEGFA, and HIF-1α, key targets enriched in the VEGF signaling pathway, with oleanolic acid was performed. It is generally accepted that an affinity of < −6 kcal mol −1 indicates a good binding effect, while < −9 kcal mol −1 indicates a very good binding effect. As seen in Table 2, oleanolic acid exhibited good affinity with PKC-α. Molecular docking by Autodock showed that the binding free energy was relatively high ( Figure  4C). Overall, these results showed that oleanolic acid had a good affinity for PKC-α.

Results of retinal electrophysiological examination
For F-ERG, a-wave ( Figure 5A) and b-wave ( Figure 5B) amplitudes were significantly lower in the DR group compared with the NC group, while the a-wave ( Figure 5D) and b-wave ( Figure 5E) latencies were higher in the DR group than in the NC group. Each treatment significantly increased the amplitude of a-and b-waves and normalized the a-and b-wave latencies. The DR group decreased significantly compared with the OPs and OS2 in the NC group. Amplitude increased in each group after treatment but remained lower than in the NC group ( Figure 5C,F).

Blood-retinal barrier (BRB) breakdown measurement
Evans Blue leakage was significantly higher in DR compared with the NC group. Moreover, the mean leakage in each treatment group was significantly reduced compared to the model group ( Figure 6).

Histopathological observation of the retina
The NC group rats exhibited a neat arrangement of retinal layers, normal ganglion cells, no edema, and atrophy. DR rats had substantially fewer retinal ganglion cells, vacuole-like changes in cells, and fewer cells in the inner nuclear layer exhibiting atrophy. Doxium treatment group rats had reduced retinal ganglion cells, vacuole-like changes of cells, neater arrangement of the inner nuclear layer, and no edema. Retinal ganglion cells of rats in the OA-L group presented   vacuolar-like changes, disorganized arrangement of the inner nuclear layer and significant edema. Retinal ganglion cells in the OA-H group exhibited vacuolar-like changes, with a neat arrangement of the inner nuclear layer and mild edema. In the ZJP group, the retinal layers were systematically organized, and the ganglion cells were vacuolated ( Figure 7A).
Compared with the NC group, the collagen fibers of the posterior hyaloid membrane were considerably proliferated in DR rats. Moreover, the proliferation of collagen fibers of the posterior hyaloid membrane was reduced in each treatment group compared with the DR group ( Figure 7B).

Immunohistochemistry of VEGFA in retinal tissue
Compared with the NC group, VEGFA expression was significantly increased in the retinal tissue of DR rats. Compared with DR, VEGFA expression in retinal tissues of rats in the Doxium group, OA-H group and ZJP group was decreased to varying degrees ( Figure 7C). In this respect, VEGFA was more concentrated in the retinal nerve fiber layer.

Detection of VEGF signaling pathway and prediction of target protein expression by Western blot
Compared with the NC group, the expression of HIF-1α, VEGFA, PKC-α, and P-ERK1/2 in the retinal tissues of DR rats was significantly higher. Moreover, the expression of HIF-1α, VEGFA, PKC-α, and P-ERKK1/2 in the retinal tissues of rats in the Doxium group, OA-H group and ZJP group exhibited different degrees of reduction (Figure 8).

Binding capacity of PKC-α to oleanolic acid
SPR binding analysis showed that the affinity constant was K D = 1.73 μM, and oleanolic acid could be fitted to the curve and showed good affinity (Figure 9).

Discussion
In the present study, our comprehensive database mining yielded 42 potential bioactive components, 1150 DR-related targets, and 63 common targets of ZJP and DR. PPI network analysis showed that oleanolic acid was the active ingredient with the highest nodal degree value and played a key role in ZJP to improve DR. PRKCA, HIF-1α, VEGFA, and MAPK were identified as significant genes based on the disease target network analysis results. It is well-established that the core of diabetic biochemical abnormalities is hyperglycemia. In hyperglycemic states, mitochondrial dysfunction causes enhanced oxidative stress and massive production of reactive oxygen species leading to altered glycolytic pathways [35], which ultimately results in increased DAG synthesis and activation of PKC-α [36][37][38]. The PKC-α family can cause retinal blood flow through multiple pathways, such as promoting extracellular matrix protein synthesis, leading to leukocyte adhesion and endothelial cell activation and proliferation altered kinetics and disrupting the blood-retinal barrier [39][40][41]. In this respect, superoxide ions and  protein hydrolases can be produced when stagnant leukocytes increase in size and adhere to vascular endothelial cells [42]. At the same time, the enlarged leukocytes have poor durability, and stagnation in the vasculature can cause impaired retinal capillary microcirculation, resulting in a lack of oxygen supply due to leaky and non-perfused microvasculature in the fundus [43][44][45]. This retinal ischemia and hypoxia state exacerbates the inflammatory response and induces upregulation of hypoxia-inducible factor HIF-1α [46]. It has been shown that HIF-1α acts as a transcription factor that regulates the expression of multiple midstream and downstream factors [47], including VEGFA and its receptors, which are most critical for angiogenesis [48], and activated PKC-α can promote increased expression of vascular endothelial cell growth factor [49]. Indeed, the essence of retinal neovascularization is to improve the state of retinal ischemia and hypoxia, but the lack of complete tissue coverage makes this new blood vessel structurally incomplete, thinner and more brittle compared to the normally repaired vessels, which can easily leak or rupture and bleed, damaging the normal retinal tissue structure leading to impaired vision.
Functional analyses showed that ZJP and its main component, oleanolic acid, were mainly enriched in the VEGF signaling pathway, whose main role is to promote angiogenesis while preserving existing vessels. However, vascular abnormalities may occur when endothelial growth factors are much higher than anti-vascular endothelial growth factors [50]. Vascular endothelial growth factor (VEGFA) is a direct and primary stimulating vascular growth factor that plays an important role in neovascularization [51]. Various factors associated with angiogenesis directly or indirectly induce the expression of VEGFA and its receptors to contribute to enhanced retinal microvascular endothelial cell permeability, further leading to the occlusion of retinal microvessels or pro-pathological revascularization. After retinal vascular endothelial cell injury, apoptosis and loss, endothelial progenitor cells (EPC) migrate to the injury site to participate in endothelial cell repair and replace the injured vascular endothelial cells [52]. Given their potential to proliferate, migrate, and differentiate into endothelial cells, EPCs participate in the repair of injured vascular endothelial cells and physiological and pathological neovascularization [53]. When VEGFA activates phosphatidylcholine-specific phospholipase C (PLC3) by binding to VEGFA receptors specifically expressed on retinal endothelial cells, it hydrolyzes phosphatidylinositol biphosphate to produce diglycerolipids (DAG) and inositol polyphosphate (IP), where DAG activates protein kinase C (PKC) in the cytosol and anchors it to the membrane. Later, the signal is transferred into the cell via the tyrosine-protein kinase pathway, activating the MAPK signaling pathway to stimulate the proliferation and migration of vascular endothelial cells and alter the extracellular matrix, which then induces endothelial cell growth, aggravates vascular permeability or promotes retinal neovascularization [54].
The molecular docking simulation technique is a convenient and effective way to investigate the interaction of small molecules with target targets. In the present study, we used Vina 1.1.2 software to perform docking studies of the compound Oleanolic acid with H1F-1α, VEGFA, PKC-α, P-ERK1, and P-ERK2, respectively. Our findings suggested that the PKC-α target may be a key target. Next, we evaluated the binding mode of Oleanolic acid and PKC-α, which formed hydrophobic interactions with ASP475 and ALA521 and hydrogen bonds with LYS409 ( Figure 4C) [55][56][57].
Electroretinography is mainly used clinically to diagnose lesions between photoreceptor cells to ganglion cells []. The a-wave is mainly derived from the receptor potential of photoreceptor cells, while the b-wave is mainly derived from the conduction of visual im-pulses by bipolar cells. Oscillatory potentials are often used to diagnose DR early [58]. The OPs wave is selectively reduced or disappears at the onset of DR, reflecting an ischemic state in the retina. In the present study, OPs wave amplitude was reduced in DR rats compared with the NC group, and OPs wave amplitude was elevated in each treatment group compared to DR. The b-wave amplitude of the electroretinogram was reduced in DR rats compared with the NC group. Compared with DR, the increase in the b-wave amplitude of the electroretinogram in each treatment group suggested that ZJP and its main component, oleanolic acid, yielded a protective effect on retinal neurological impairment in diabetic rats. Moreover, we found that the EB leakage in the DR group and each treatment group was significantly higher than in the NC group, indicating that the blood-retinal barrier was damaged, and leakage increased during the early stage of diabetes mellitus in rats. EB leakage in each treatment group was reduced compared with the DR group, suggesting that ZJP and its main component, oleanolic acid, could reduce the retinal leakage caused by hyperglycemia, thus maintaining the integrity of the diabetic blood-retinal barrier.
Other related studies found that ZJP could improve weight loss in diabetic rats and inhibit the overexpression of ICAM-1, TNF-α, and IL-1β in retinal tissue [59]. Among them, modern pharmacological studies have found that Lycium barbarum has antioxidant and anti-apoptotic effects on retinal photoreceptor cells and has neuroprotective effects [60]. There is a growing consensus that Cuscuta, Broussonetia and Cyperus have antioxidant properties beneficial in treating neurodegenerative diseases and inhibit retinal photoreceptor cell apoptosis [61][62][63]. Moreover, oleanolic acid and its derivatives have anti-inflammatory and hypoglycemic effects and mitigate tissue edema and ischemia-reperfusion injury [64][65][66][67]. Our experimental results substantiated that ZJP and its main component, oleanolic acid, could inhibit the proliferation of collagen fibrous tissue and improve the histopathological changes in the retina of diabetic rats.

Conclusions
In this study, we predicted the effect and mechanism of the ZJP on DR through network pharmacology prediction analysis, molecular docking and animal experiments. Network pharmacology showed that the ZJP might improve DR through multiple targets and pathways. Oleanolic acid was found to be the main active ingredient. Molecular docking showed that oleanolic acid had the most significant potential activity, similar to PKC-α. Animal experiments substantiated that oleanolic acid in ZJP and its main components might play an important role in inhibiting the VEGF