The prevalence of patients with type 2 diabetes mellitus (T2DM) has significantly increased with the progress of human living standards [1]. T2DM is the predominant form of diabetes around the world compared to other types of diabetes which account for around 90% of diabetes cases and it poses a substantial social burden. The hallmark features of T2DM patients include impaired insulin action and abnormal secretion, which can lead to various complications, such as retinopathy, peripheral neuropathy, macro vascular diseases, microalbuminuria, nerve pain, fibromyalgia, and anxiety, et al. Currently, metformin is the first-line medication to treat T2DM [2,3]. Research has demonstrated that metformin alleviates hyperglycemia-induced endothelial impairment by downregulating autophagy via the Hedgehog (HHS) pathway [4], and metformin can treat diabetes by modulating HHS-mediated autophagy [5]. However, the use of metformin is limited due to its potential toxic side effects [6].

Diabetic retinopathy (DR), a major complication of T2DM, is a leading cause of vision loss worldwide [7]. As the prevalence of diabetes increases, the incidence of DR is also expected to increase [8]. Many patients are suffering from decreased vision or even blindness due to the DR [9]. Therefore, there is an urgent need for affordable and effective treatments to combat DR and enhance the quality of life for diabetic patients.

Astragalus polysaccharide (APS) is a bioactive component of the leguminous plant Astragalus mongholicus. Widely used in traditional Chinese medicine, APS has been shown to protect the liver and kidney [10,11], enhance immune function [12], and mitigate hypertensive [13], oxidative stress [14], and bacterial infections [15]. Previous behavioral studies have also demonstrated that APS could improve insulin sensitivity and protect renal function of diabetic patients [16]. However, the effects of APS on the treatment of DR remain largely unexplored. Therefore, the purpose of this study is to investigate the effects of APS on DR and to explore its underlying mechanisms.

Network pharmacology is an emerging interdisciplinary discipline that integrates pharmacology, systems biology, biomedicine, and network biology [17]. In this study, the anti-T2DM targets of APS and the binding activity between APS and T2DM-related targets were explored through network pharmacology analysis, and the corresponding Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and gene ontology (GO) terms of APS’s anti-T2DM targets were found through a series of network pharmacology tools and websites.

Aquaporins (AQPs) are a family of integral membrane proteins that facilitate the transport of water cross the plasma membrane [18]. They play a crucial role in maintaining ionic and osmotic balance in the central nervous system by responding to osmotic gradients and differences in hydrostatic pressure [6]. As a member of the AQPs family, AQP1 is typically expressed in the distinct amacrine cells of the outer retina [19]. It has been shown that the development of edema is associated with an alteration in glial water channel expression in the brain [8]. An increased expression of AQP1 in the reactive glial cells was observed in various edema-related neurological disorders such as ischemia [20], trauma [21], subarachnoid hemorrhage [22], and brain tumors [23].

Iandiev et al. [24] confirmed that transient ischemia causes an increase in immunoreactive AQP1 in retinal glial cells. Kaur et al. [25] demonstrated that the upregulation of AQP1 enhances the invasive potential of glioma cells [25,26]. Khoshnoud et al. [27] showed that the intervention of APS could inhibit the expression of AQP2 in the kidney of diabetic rats, thus protecting the kidney. The regulatory relationship between SHH, Gli1, and AQP1 has been established by several studies [28,29]. Additionally, our previous studies showed that the expression of AQP2 in the model of diabetic rats was excessively increased [30]. However, it is not known whether DR is associated with the expression changes of AQP1. Therefore, building on prior studies concerning AQP2, we aim to investigate the expression of AQP1 in the retinas of diabetic rats and explore the relationship between the SHH-Gli1-AQP1 pathway and retinal reconstruction.

APS target fishing and T2DM-associated target screening

The flowchart of network pharmacological analysis is illustrated in Fig. 1.

Figure 1. Workflow for screening target genes using the network pharmacology approach. APS, astragalus polysaccharide; T2DM, type 2 diabetes mellitus; KEGG, Kyoto Encyclopedia of Genes and Genomes; GO, gene ontology; PPI, protein-protein interaction.

APS targets were predicted using the Swiss Target Prediction database (http://www.swisstargetprediction.ch/) and the BATMAN-TCM database (http://bionet.ncpsb.org/batman-tcm/). The T2DM-associated targets were obtained from the GeneCards database (https://www.genecards.org/) and the DisGeNET database (https://www.disgenet.org/).

All of the APS and T2DM target names were transformed into the official gene symbol by using the DAVID database (https://david.ncifcrf.gov/conversion.jsp/). The ‘target genes’ of the APS’s anti-T2DM targets were established by comparing and analyzing the targets common between the T2DM-associated targets and the predicted APS targets through using the Venn online tool (https://bioinfogp.cnb.csic.es/tools/venny/index.html).

Pathway enrichment analysis (KEGG) and GO enrichment analysis of target genes

To investigate the potential therapeutic mechanisms of target genes, we conducted GO and KEGG pathway enrichment analyses using the Metascape database and the Xiantao academic website (https://www.xiantao.love/). The Metascape database (http://metascape.org/) is a web-based portal that provides a comprehensive gene list annotation and acts as an analysis resource for experimental biologists. Metascape database combines functional enrichment, interactome analysis, gene annotation, and membership search to leverage over 40 independent knowledge bases within one integrated portal [31]. In our study, Metascape was utilized for GO analysis, including biological processes, cellular components, and molecular functions, as well as for KEGG pathway analysis. Additionally, the Xiantao academic website was employed to perform the same analyses to validate our results.

Construction of protein-protein interaction (PPI) network and subnetwork identification

Initially, the APS’s anti-T2DM targets were imported into the online database STRING (https://string-db.org/) for PPI analysis for observing the interactions between target genes from subcellular proteomics profiling. In this study, the confidence level was set to larger than 0.4 to construct the PPI network and the PPI network was subsequently visualized with Cytoscape 3.7.2.

Molecular docking simulation of Gli1 and AQP1 with APS and metformin

Molecular docking analysis predicts the interaction strength and binding modes between macromolecular proteins and small molecules [32]. Atomic-level three-dimensional (3D) structure of the target genes was searched in the RCSB Protein Data Bank database (http://www.rcsb.org/pdb/home/home.do), which is an important database containing the atomic-level 3D structural data of most biological macromolecules [33]. Then the 3D chemical structure of APS was obtained by the ZINC database (http://zinc.docking.org/). Using Autodock software, molecular docking simulations were conducted to analyze the interactions of Gli1 and AQP1 with APS and metformin, determining their respective free binding energies and binding modes [34].

Ethics statement

All the procedures about animal experiments were performed in compliance with the Institute’s guidelines and the Guide for the Care and Use of Laboratory Animals. The study was approved by the institutional animal care committee of Weifang Medical University (NO. WFMU-2022-0301).

Animals

Male eight-week-old Wistar rats (200–220 g) were sourced from the Laboratory Animal Center of Weifang Medical University. Prior to the experiment, the rats were acclimatized to room temperature, provided a normal diet, and kept the conditions of light and dark circulation. Blood glucose testing and intragastric administration have proceeded during the light phase of the circulation. Fifty healthy male Wistar rats were randomly divided into the normal group (Con group, n = 10) and the model group (DM group, n = 40). Eighteen weeks after the model group was fed a high-sugar and high-fat diet, the streptozotocin (STZ) solution was intraperitoneally injected at a dose of 25 mg/kg for 2–4 consecutive days. 72 h later, the randomized blood glucose was measured. The rats with random blood glucose ≥ 16.7 mmol/L were seen as successful models and were randomly divided into the DM group (n = 10), APS high-dose group (n = 10), APS low-dose group (n = 10), and metformin group (n = 10). The APS were purchased from Shanghai Yuanye Biotechnology Co., Ltd. (MDL: MFCD02180551, MW: 254.69).

The rats in the APS low-dose group and APS high-dose group were intragastrically administered with APS for 200 mg/kg/D and 400 mg/kg/D, respectively, and the metformin group was intragastrically administered with metformin for 250 mg/kg/D by gavage daily for 12 weeks. The DM group and control group were given the same volume of normal saline at the same time. The fasting blood glucose was measured once a week during the experiment. At the conclusion of the experiment, the blood samples was collected from the eyeballs of rats, and the retinal tissue was taken out for subsequent analysis.

Morphological experiments

The hematoxylin and eosin (H&E) staining was performed in each group of rat retinal sections to determine the cell structure changes and sugar content changes caused by DR in rats. The rat eyeballs were embedded in the paraffin and were cut into 5 μm tissue pieces, spread out in warm water at 45°C, attached, and baked at 60°C for 4 h. These sections were immersed in xylene and gradient ethanol for conventional dewaxing and rehydration. Subsequently, the sections were stained with hematoxylin solution for 5 min at room temperature, rinsed with running water, and soaked in hydrochloric acid ethanol solution for several seconds. After rinsing, they were reacted with an eosin staining solution for 5 sec, followed by dehydration, and transparency. Finally, the sections were sealed with a neutral gum and observed under the microscope.

The Periodic Acid-Schiff (PAS) stain was performed to determine the sugar content in the retinal cells of each group of rats. The rat eyeballs were embedded in paraffin, cut into 5 μm tissue pieces, spread out in warm water at 45°C, attached, and baked at 60°C for 4 h. These sections were immersed in xylene and gradient ethanol for conventional dewaxing and rehydration. After this, the sections were stained with periodic acid for 10 min at room temperature, and rinsed with running water, after rinsing, they were reacted with a Schiff staining solution for 10 min, stained with hematoxylin for 5 min, followed by dehydration, transparent. The sections were finally sealed with a neutral gum and observed under the microscope.

Immunohistochemistry

To determine the expression levels of AQP1, SHH, and Gli1 in the retina of rats, immunohistochemical staining was performed in the retinal sections.

Briefly, the rat eyeballs were embedded in paraffin, cut into 5 μm tissue pieces, spread out in warm water at 45°C, attached, and baked at 60°C for 4 h. Then the sections were immersed in xylene and gradient ethanol for conventional dewaxing and rehydration. Subsequently, the sections were immersed in citrate buffer and repeatedly heated using a microwave oven for 5 min to remove the antigen. After rinsing, an endogenous peroxidase blocker was added dropwise to the tissue sections, reacted for 10 min, rinsed, and sealed in goat serum for 20 min. The goat serum sealing solution was discarded, and the resulting sections were incubated with anti-AQP1 and SHH, Gli1 primary antibody (1:200) in a refrigerator at 4°C overnight. On the next day, the washed sections were added dropwise with a secondary antibody solution, reacted at 37°C for 10 min, and rinsed thoroughly, then reacted with diaminobenzidine. Finally, the sections were counterstained with hematoxylin, sealed, and observed.

Cell culture

Rat Müller cells (MIO-M1; YB-H3309, Ybio) were cultured in Dulbecco’s Modified Eagle’s Medium complete medium containing 10% fetal bovine serum (10,566,016, Gibco), 5 mg/ml streptomycin, and 5 U/ml penicillin (2211093, Gibco) at 37°C with 5% CO₂. The cell density was close to 80% for digestion and passage. The MIO-M1 cells cultured with 5 mM glucose for 12 h were taken as the control group, while MIO-M1 cells cultured with 35 mM glucose for 12 h were regarded as the high-glucose groups.

MIO-M1 cell transfection experiments

Small interfering RNA (siRNA) and negative control of AQP1, and Gli1 were synthesized by Gene Pharma Biotech Co., Ltd. Overexpression plasmid of SHH was constructed in GeneSeed Co., Ltd. MIO-M1 cells were transfected with Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) following the manufacturer's instructions. RNA-related indexes were detected 24 h after the transfection. Protein-related indexes were assessed 48 h after the transfection.

Western blotting analysis

Retinal tissue homogenate were processed using RIPA lysis buffer (Solarbio) containing 50 mM Tris (pH 7.4), 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate and 1 mM phenylmethylsulphonyl fluoride. Then the lysates were centrifuged at 4°C at 12,000 rpm for 10 min and protein concentrations were analyzed by the BCA method (CWBIO). After protein electrophoresis and transblotting, the blots were blocked with 5% skimmed milk for 1 h and then probed with primary antibody against STAT (1:2,000, Proteintech), VEGF (1:2,000, Proteintech), AQP1 (1:1,000, Proteintech), SHH, Gli1 (1:1,000, Proteintech), and β-actin (1:5,000, Proteintech) overnight at 4°C. Then, the membranes were incubated with the secondary antibody (1:5,000) for 1 h at room temperature. Detection of antigen-antibody complexes was performed using enhanced chemiluminescence reagent (Millipore), and band densities were quantified using Image J software.

ELISA kit assay

Cell lysates from Retinal Müller cells and retinal tissue homogenates from rats were collected to assess the expression levels of AQP1, SHH, and Gli1 using ELISA kits, following the manufacturers’ instructions.

Quantitative real time-PCR

Total RNA from retinal tissue homogenates was extracted by using Trizol reagent (Invitrogen) and reversely transcribed using the reverse transcription kit (TOYOBO FSQ-101). Quantitative real time-PCR was performed by using a QuantiFast SYBR Green PCR Kit (QIAGEN). Sets of PCR primers: AQP1: forward 5’- CCCTCTTCGTCTTCATCAGC-3’ and reverse 5’-CTGAGCCACCTAAGTCTCGG-3’; SHH: forward 5’-TTAAATGCCTTGGCCATCTC-3’ and reverse 5’-TTTCACAGAGCAGTGGATGC-3’; Gli1: forward 5’-CTCTGCTGACTCTGGGATATG-3’ and reverse 5’-GATCAGGATAGGAGCCTG-3’; β-actin: forward 5’-TCACCCACACTGTGCCCCATCTACGA-3’ and reverse 5’ CAGCGGAACCGCTCATTGCCAATGG-3’. The primers were designed with reference to the National Centre for Biotechnology Information (NCBI) database of conserved coding regions. The PCR was performed in Light-Cycler 480 (Roche) PCR machine using the following cycle parameters: 1 cycle of 95°C for 5 min, and 40 cycles of 95°C for 10 sec, 60°C for 30 sec. A comparative threshold cycle (CT) method was used, and the PCR data were calculated using the 2^–ΔΔCT method.

Immunofluorescence assay

MIO-M1 cells were seeded onto 12-well plates and transfected with siRNA or plasmid. The expression of STAT and VEGF were assessed post-transfection. Cells were firstly immobilized with 4% paraformaldehyde for 25 min, followed by 0.1% TritonX-100 permeable cells for 15 min. Then the cells were rinsed in phosphate-buffered saline (PBS) three times and then blocked with 1% bovine serum albumin for 1 h. The above operation is conducted at room temperature. Then diluted primary antibody (1:1,000) was added to the well plate and incubated at 4°C overnight. The cells were rinsed with PBS three times and further incubated with fluorescent secondary antibody at room temperature for 1 h in darkness. Cell nuclei were stained with DAPI solution at 37°C for 15 min, then the cells were rinsed with PBS three times. After the anti-fluorescence quenching sealing tablets were dropped and the cover glasses were sealed, the cells were observed and photographed under a fluorescence microscope.

Cellular thermal shift assay (CETSA)

CETSA can be used to understand the thermal stability of proteins upon binding to ligands. Briefly, the cells were heated under different temperature (20°C–80°C) after treating with APS or solvent control; then the thermal stability of target proteins were detected via Western blot assay.

Statistical analysis

All of the data in these studies were expressed as mean ± standard deviation. The student’s t-test or one-way analysis of variance was used to compare differential significance with control value using SPSS 13.0 software. A p-value of less than 0.05 was considered to be statistically significant.

Screening and identification of APS-T2DM-related targets

As shown in Fig. 2A, a combination of the BATMAN-TCM and Swiss Target Prediction was intended to predict the putative targets of APS and 502 targets were collected through searching these databases. The DisGeNET databases and GeneCards were searched and a total of 1,001 T2DM-related targets were identified after removing the duplicate results. To determine the association between the APS potential targets and T2DM-related target genes, the Venn online tool (https://bioinfogp.cnb.csic.es/tools/venny/index.html) was used and 76 shared targets between APS and T2DM were obtained.

Figure 2. Results of network-pharmacology analysis. (A) Obtain of APS-T2DM-related targets. (B) PPI network of APS-T2DM related targets. (C) Sub-network of the PPI network. (D) Enrichment analysis results of targets by Metascape. (E, F) Enrichment analysis results of targets by Xiantao academic website. (G) qPCR test of 10 targets. (H) 3D structures and binding sites of ligands and receptors used for molecular docking between the AQP1, GLI1, and the APS, metformin; and the molecular docking energy score (I). (J–M) Cellular thermal shift assays (CETSA) indicated APS can improve thermostability of AQP1 and GIL1. APS, astragalus polysaccharide; T2DM, type 2 diabetes mellitus; PPI, protein-protein interactions.

PPI network construction and analysis discovered the potential relationship between T2DM and Gli1/AQP1

In the ‘Screening and identification of APS-T2DM related targets’ section, the T2DM targets and APS targets showed 76 duplicate genes. It suggested that these genes might be the targets of APS for the treatment of T2DM. To study the interaction of the targets and search for the hub genes, the PPI network analysis of the potential targets was then conducted (Fig. 2B). In addition, the PPI subnetwork was also produced by using the CytoHubba plug-in in the Cytoscape software. As shown in Fig. 2C, the 10 predicted targets (hub genes) which include the Gli1 and AQP1 were grouped in this sub-network. Moreover, RNA analysis of these 10 targets revealed significantly elevated expression levels of Gli1 and AQP1 in the diabetic model rats (Fig. 2G).

GO and KEGG pathway enrichment analysis of target genes discovered the potential relationships between APS and DR

To clarify the characteristics of the 76 targets and to illustrate the mechanism underlying the effects of APS on T2DM treatment more comprehensively, we performed the GO and KEGG enrichment analysis by Metascape (Fig. 2D) and Xiantao academic website (Fig. 2E, F).

The result showed that the APS-T2DM related targets are involved with the DR-related items which include the ‘response to hypoxia’, ‘blood circulation, ‘NO/cGMP/PKG mediated Neuroprotection’, ‘Synaptic Transmission’, and ‘Regulation of Neurotransmitter Levels’, et al. Which indicated that APS may have effects on DR.

Molecular docking analysis and CETSA proved the affinity between APS and AQP1/GLI1

The AQP1 and GLI1 were selected for molecular docking analysis with APS and metformin (Fig. 2H). As shown in the heat map of Fig. 2I, the binding energy of APS with AQP1 and GLI1 was lower than the binding energy of metformin with AQP1 and GLI1. Besides, CETSA indicated APS can improve thermostability of AQP1 and GIL1, this results further prove that APS might binding to AQP1 and GLI1 (Fig. 2J–M). The above results suggest that APS may exert therapeutic effects in T2DM by targeting AQP1 and GLI1.

Anti-DR effects of APS

Fig. 3A–C illustrates the evaluation of body weights, insulin levels, and oral glucose tolerance test results, indicating the anti-diabetic effects of APS.

Figure 3. Western blot analysis indicated the anti-DR effects of APS. (A) Body weight change of rats during experiment. (B) Insulin levels in different groups (^##p < 0.01 compared with the control group; *p < 0.05 and **p < 0.01 compared with the model group). (C) Oral glucose tolerance test (OGTT) of rats during experiment. (D–F) Two DR relative indicators VEGF and STAT were reduced by APS. n = 3 in each group, values are presented as mean ± SD. ^#p < 0.05 compared with the control group; *p < 0.05 compared with the model group. DR, diabetic retinopathy; APS, astragalus polysaccharide.

To verify the anti-DR effect of APS, we detected two DR related markers STAT and VEGF (Fig. 3D–F). The results indicate that the expression of STAT and VEGF were significantly increased; however, after APS treatment, their expression levels were decreased. These results indicated the anti-DR effects of APS. Besides, as shown in Fig. 4A, the morphology of the rat retinal was observed via H&E staining. The normal group exhibited regularly arranged retinal cells with normal morphology, round orderly-arranged ganglion cells, and normally stained and clear cell nuclei without evident damage, while the model group had disorderedly arranged retinal cells with changed morphology, some of which were loosely arranged and poorly stained; the cells showed nuclear fragmentation and even disappearance, as well as more visible cavitations in some of them. However, compared with the model group, the retinal cells of the metformin group and the APS group were arranged neatly and the morphology was improved. The PAS staining was used to visualize the glycogen and other polysaccharides. What is more, the morphology of the retina observed by PAS staining is shown in Fig. 4B. The retinal cells in the normal group are regularly arranged and present normal morphology. As for the model group, the retinal arrangement is disordered, and the glycogen staining is significantly deeper than in the normal group. However, compared with the model group, the retinal cell coloration of the metformin group and the APS group was relatively weakened. These results collectively underscore the anti-DR effects of APS.

Figure 4. Morphology detection of the rat retina tissues demonstrated the association between the SHH-Gli1-AQP1 pathway and DR. (A) Retinal morphology was observed via H&E stain. (B) Retinal morphology was observed via PAS staining. Immunohistochemistry detection of (C) AQP1, (D) Gli1, and (E) SHH. The magnifications are 400×. DR, diabetic retinopathy; T2DM, type 2 diabetes mellitus; APS, astragalus polysaccharide; SHH, Sonic Hedgehog.

Histological experiments demonstrated the association between the SHH-Gli1-AQP1 pathway and DR

To detect the expression of the AQP1, Gli1, and SHH protein levels in the retinal cells, immunohistochemistry stains were conducted. As shown in Fig. 4C, the positive staining area of AQP1 was increased in the model group compared with the normal group. However, after treatment with the APS or metformin, the expression of AQP1 was significantly reduced. Similarly, the expression of Gli1 (Fig. 4D) and SHH (Fig. 4E) were also increased in the model group and were reduced after treatment with the APS or metformin.

These findings demonstrate that DR involves the SHH-Gli1-AQP1 pathway, and APS effectively mitigates DR by targeting this pathway.

Western blotting and RT-qPCR detection proved the association between the SHH-Gli1-AQP1 pathway and DR

As depicted in Fig. 5A–C, the mRNA expression levels of AQP1, Gli1, and SHH were significantly elevated in the model group compared to the normal group. However, as for the metformin group and the APS group, the mRNA expression levels of AQP1, Gli1, and SHH were lower than that of the model group. As shown in Fig. 5D–G, the expression levels of AQP1, Gli1 and SHH protein in the model group were significantly higher than those in the normal group. However, the relative protein expression levels of AQP1, Gli1, and SHH in the metformin group and the APS group were decreased in a dose-dependent fashion when compared with the model group. These results further proved the association between DR and the SHH-Gli1-AQP1 pathway and APS could restrain the DR through this pathway.

Figure 5. RT-qPCR and Western blotting analysis proved the association between the SHH-Gli1-AQP1 pathway and DR. (A–C) mRNA levels detection of AQP1, Gli1, and SHH by RT-qPCR (n = 6). (D–G) Protein levels detection of AQP1, Gli1, and SHH by Western blotting (n = 3). ^##p < 0.01 compared with the control group; *p < 0.05 and **p < 0.01 compared with the model group. Values are presented as mean ± SD. DR, diabetic retinopathy; APS, astragalus polysaccharide; SHH, Sonic Hedgehog.

Immunofluorescence assay and rescue analysis of MIO-M1 cells demonstrated the association between the SHH-Gli1-AQP1 pathway and DR

As depicted in Fig. 6A, compared to the control group, the fluorescence levels of STAT and VEGF in the model group were significantly increased, However, as for the metformin group and the APS group, the fluorescence levels of STAT and VEGF were lower than that of the model group. As shown in Fig. 6B, C, the expression levels of STAT and VEGF protein in the model group were significantly higher than those in the control group. Following treatment with si-SHH, the relative protein expression levels of STAT and VEGF were decreased when compared with the model group. However, when the Gli-1 was overexpressed, the effect of si-SHH was rescued. Meanwhile, overexpression of Gli-1 could also rescue the effect of si-AQP1 (Fig. 6D, E). These results further proved the association between DR and the SHH-Gli1-AQP1 pathway.

Figure 6. Rescue analysis proved the association between the SHH-Gli1-AQP1 pathway and DR. (A) Detection of DR related proteins STAT and VEGF by immunofluorescence stain. The magnifications are 200×. (B–D) Overexpression of Gli-1 and knockdown of SHH/AQP1 indicated the association between the SHH-Gli1-AQP1 pathway and DR. (E) Knockdown of AQP1 decreased the VEGF protein level compared with the model group, but over-expression of Gli-1 inhibited this effect (^##p < 0.01 compared with the control group; *p < 0.05 compared with the model group, ^&p < 0.05 compared with the si-AQP1 group). n = 3 in each group, values are presented as mean ± SD. ^##p < 0.01 compared with the control group; *p < 0.05 and **p < 0.01 compared with the model group, ^&p < 0.05 and ^&&p < 0.01 compared with the OE Gli-1 group. DR, diabetic retinopathy; APS, astragalus polysaccharide; SHH, Sonic Hedgehog.

DR represents a critical complication of diabetes, capable of causing severe and often irreversible vision impairment, and stands as a leading cause of blindness globally. With the increasing prevalence of diabetes in society due to improved living standards, the incidence of DR has also risen sharply. Understanding the pathogenesis of DR and developing effective treatments are urgent priorities. There has a complex developmental mechanism and pathophysiology of DR; Barber [35] has shown that chronic hyperglycemia can affect the whole retinal components, and lead to neuronal cell degeneration, glial dysfunction, and even microvascular damage. Besides, Kowluru [36] and Devi et al. [37] have shown that retinal edema, vascular proliferation, microvascular pericyte injury, and the müller cell injury play important roles in the progression of DR.

AQP1 is a protein typically expressed in the outer retina photoreceptors which were formed from the amacrine cells, retinal pigmentary cells, and astrocytes. The abnormal expression of AQP1 often occurs under pathological conditions such as retinal ischemia and DR (especially around the superficial blood vessels of the retina). One of the key steps of the development of DR is the formation of new blood vessels which requires the proliferation, adhesion, and metastasis of endothelial cells. Here, AQP1 plays a significant role in the basic stage of neovascularization during the process of endothelial cell metastasis. This underscores the significant role of AQP1 in the pathogenesis of DR.

The SHH signal transduction pathway is a signaling pathway that plays a crucial role in the development of the central nervous system of vertebrates. Excessive activation of the SHH pathway is a worthy cause of tumor necrosis and malignant biological behavior maintenance in various nervous systems. Gli1 is the ultimate responder and function performer of the SHH signal pathway. SHH signaling pathway and AQP1 play a key role in the development of the central nervous system and maintenance of normal function. Studies have shown that both of them have important effects on the development of central nervous system glioblastoma. It has shown that there is a significant correlation between the high expression of Gli1 induced by the SHH pathway activation and the high expression of AQP1 [28].

In this study, we utilized network pharmacological methods to predict APS targets for treating T2DM, identifying AQP1 and Gli1 as potentially involved. Subsequently, we evaluated the expression changes of AQP1, Gli1, and explored the related SHH-Gli1-AQP1 signaling pathway in the following experiments.

Our in vivo study showed that the mRNA and protein levels of AQP1, Gli1, and SHH were aberrantly highly expressed in the retinal tissues of the DR model rats compared with the normal rats. Meanwhile, the results of the present analysis revealed that the metformin and APS could remarkably weaken the abnormally high expression of AQP1, Gli1, and SHH in DR, and can improve the retinal cell morphology of T2DM rats with DR. In vitro studies using MIO-M1 cells treated with si-AQP1/si-SHH and OE-Gli-1 further proved the relationship between DR and SHH-Gli1-AQP1 signaling pathway. Therefore, it can be concluded that APS may treat DR by affecting the SHH-Gli1-AQP1 signaling pathway. The results shown in Fig. 5 show that APS treatment reduced the expression of SHH at both mRNA and protein levels, we think that this phenomenon can be explained by the existence of feedback loops in the SHH-Gli1-AQP1 axis. Studies have shown that SHH be feed-regulated [6,7]. In our results, we think that phenomenon is because the down-regulate of Gli1 and AQP1 induced by APS feeds into a decline in the transcription of the SHH gene in a feedback manner, and hence its protein level.

This study indicated that APS may treat DR by affecting the SHH-Gli-AQP1 signaling pathway in T2DM rats. These findings propose a novel therapeutic strategy for managing DR.

This work was supported in part by grants from the National Natural Science Foundation of China (81360361).

The author would like to thank the members of the research group for their hard work and the guidance of Xiaomei An.

The authors declare no conflicts of interest.