Tangwang formula ameliorates diabetic retinopathy by inhibiting cell apoptosis and regulating the intestinal microbiota and metabolic profiles

Abstract

BACKGROUND

Tangwang formula (TWF), a traditional Chinese medicine, has been shown to delay the progression of diabetic retinopathy (DR) over the past 20 years. However, the potential therapeutic mechanisms and effective components of TWF remain unclear.

AIM

To investigate the effective components of TWF and elucidate the mechanism underlying TWF treatment for DR.

METHODS

The chemical ingredients of TWF were detected using high-pressure liquid chromatography. Network pharmacology and molecular docking were applied to identify the targets and pathways associated with TWF and DR. A DR mouse model was established by streptozotocin to evaluate the therapeutic effect of TWF in vivo. High glucose treatment was used to induce injury in mouse retinal endothelial cells (mRECs) to verify the mechanism of TWF in vitro. The 16S ribosomal RNA sequencing and untargeted metabolomics were performed to detect intestinal bacteria and fecal metabolites in DR mice.

RESULTS

Eight important active constituents were identified by high-pressure liquid chromatography. Network pharmacology results showed that Bcl2 and Casp3 were key targets between TWF and DR. Isorhamnetin-3-O-neohespeidoside, naringenin, ononin, and calycosin-7-O-β-D-glucoside from TWF had strong affinities with Bcl2 and Casp3. TWF significantly reduced random blood glucose, inhibited the levels of proinflammatory factors in DR mice and mRECs, and alleviated retinal damage by downregulating the expressions of vascular endothelial growth factor and receptor advanced glycation end products, and upregulating the expressions of zonula occludens-1 and RBP-3. TWF inhibited the apoptosis in mouse retinal tissues and mRECs. Analysis of intestinal bacteria and metabolites revealed that TWF increased the richness and evenness of intestinal microbiota and improved fecal metabolic profiles in DR mice. Importantly, seven genus bacteria were closely correlated with amino acids, fatty acids, glycerophosphocholine, and bile acid, which could participate in ameliorating retinal injury.

CONCLUSION

These findings suggest that TWF may ameliorate retinal damage in DR mice via anti-inflammatory and anti-apoptotic activities, which may be associated with modulation of the intestinal microbiota and metabolic profiles.

Key Words: Tangwang formula; Diabetic retinopathy; Cell apoptosis; Intestinal bacteria; Metabolic profiles

Core Tip: This study identified the main active components of Tangwang formula (TWF) and explored its protective mechanism against diabetic retinopathy mice induced by streptozotocin using network pharmacology, molecular docking, 16S ribosomal RNA sequencing, and metabolomics analysis. Four key active components were identified in TWF that participated in regulating the inflammation and apoptosis. TWF might reduce blood glucose, and alleviate retinal injury by modulating the intestinal microbiota and metabolic homeostasis. These findings provide a theoretical basis for the clinical application of TWF in the treatment of diabetic retinopathy.

INTRODUCTION

Diabetic retinopathy (DR) is microvascular complication of diabetes mellitus (DM) that leads to vision loss and even blindness, seriously affecting patients’ quality of life[1]. Total 233 million people in China are diabetic (more than 95% is type 2 diabetes) by 2023. The number of patients with diabetes has been increasing annually and is developing at a younger age[2]. The incidence of DR in DM patients with a disease duration of more than ten years is more than 60%[3]. At present, anti-vascular endothelial growth factor (VEGF) is one of the most frequently used treatments for DR; however, it fails to prevent the occurrence of DR in the early stage of DM[4]. Therefore, early intervention with drugs in DM is essential to delay the occurrence and progression of DR.

Chinese medicines have some advantages in the treatment of DM with complications, which possess comprehensive characteristics of mild and sustained effects, fewer side effects, multi-targeting, and multi-pathway regulation. Tangwang formula (TWF), also known as Yiqi Wenyang Formula, has been widely used clinically for 20 years at the Guang’anmen Hospital in China[5]. Clinical studies have demonstrated that TWF can effectively improve microangiopathy and delay progression in 192 patients with DR[5]. TWF has the function of tonifying Qi and resolving stasis, which contains five Chinese herbs: (1) Astragali Radix [Bunge (AMB)]; (2) Lycii Fructus [Lycium barbarum L. (LBL)]; (3) Eriocauli Flos [Eriocaulon buergerianum Koern. (EBK)]; (4) Cinnamomi Ramulus [Cinnamomum cassia Presl (CCP)]; and (5) Typhae Pollen [Typha angustifolia L. (TAL)][6,7]. Jo et al[8] reported that AMB-CCP herb-pair is used in the treatment of diabetes. AMB is the most effective Qi tonifying herbs[9], that can promote the proliferation of rat retinal microvascular pericytes, improve the blood circulation of the lesion site and maintain the structural integrity of the blood-retinal barrier (BRB) structure[10]. Importantly, AMB significantly reduces the fasting blood glucose (Glu)[11]. LBL, EBK, CCP, and TAL are often used to treat ophthalmic diseases. LBL has the function of replenishing Qi and nourishing blood, nourishing the liver to improve visual acuity[12,13]. EBK has the ability to improve visual acuity and remove nebula to prevent cataracts[14]. CCP has the function of warming and activating meridians, promoting Yang and transforming Qi[15]. TAL promotes blood circulation to treat the fundus hemorrhage[16]. Recent studies have demonstrated that inflammation, oxidative stress, and intestinal microbiota dysbiosis are implicated in the pathogenesis of DR[17,18]. Inflammation is the main pathological factor of DR, and some inflammatory cytokines and adhesion molecules damage retinal endothelial cells (RECs) and disrupt the BRB, ultimately causing retinal microvascular leakage and retinal hemorrhage[19]. Therefore, inhibition of inflammation is an important strategy for alleviating DR development. TWF has strong anti-inflammatory and anti-oxidant effects that delay DR formation by regulating the PRC2/p38 mitogen-activated protein kinases-related pathway[20]. Calcium dobesilate (CaD), an antioxidant and a vascular protective agent is used for clinical treatment of DR by inhibiting the inflammatory factor release and eliminating free radical damage[21,22]. CaD can improve diabetic endothelial dysfunction, inhibit the apoptosis of vascular cell[23]. Accordingly, CaD was employed as a positive control in this study.

RECs play important roles in retinal vascular homeostasis. In the early stages of DR, high Glu causes the release of inflammatory factors via the nuclear factor kappa B signaling pathway[24] and promotes the interactions between the advanced glycation end products (AGEs) and receptor AGEs (RAGE) on the cell surface, and thereby further triggering retinal cell apoptosis and increasing VEGF expression[25]. High expression of VEGF in the retina promotes pathological neovascularization, whereas low expression of zonula occludens-1 (ZO-1) causes changes in vascular permeability and microstructure[26]. Microbiota dysbiosis also causes low-grade, local and systemic inflammation[27] and indirectly damages multiple organs by disturbing the glycolipid metabolism and immune homeostasis[28]. Previous studies have shown that TWF has preventive and therapeutic effects on DR by suppressing inflammation and oxidative stress[5,6]. However, the underlying mechanisms of TWF on modulating intestinal microbiota and metabolic profiles have rarely been investigated.

In this study, we mainly studied the chemical ingredients of TWF using high-pressure liquid chromatography (HPLC), the therapeutic targets of TWF on DR mice were identified using network pharmacology and molecular docking. Furthermore, the therapeutic effects and mechanisms of TWF on DR mice and mouse RECs (mRECs) were elucidated in vivo and in vitro using 16S ribosomal RNA (rRNA) gene sequencing and untargeted metabolomics. Our results will supply the theoretical basis for the use of TWF in clinical treatment of DR.

MATERIALS AND METHODS

Chemicals and reagents

TWF was supplied by the Guang’anmen Hospital at the Chinese Academy of Chinese Medicine. The mREC line was obtained from Otwo Biotech (lot No. HTX2597, China). Streptozotocin (STZ) (lot No. S8050), and CaD (lot No. IC2280) were supplied by Solarbio Life Sciences (Beijing, China). Glu (lot No. 020586) and homocysteine (HCY) (lot No. 024882) were purchased from Mindray Animal Care (Shanghai, China). Interleukin-1β (IL-1β) (lot No. 02323M2), interleukin-6 (IL-6) (lot No. 02446M2), and tumor necrosis factor-α (TNF-α) (lot No. 02415M2-JK) enzyme-linked immunosorbent assay kits were purchased from Jiangsu Jingmei Biological Technology Co., Ltd. (China). IL-1β antibody (lot No. 12742), IL-6 antibody (lot No. 57315), TNF-α antibody (lot No. 52746), VEGF antibody (lot No. 53463), ZO-1 antibody (lot No. 33725), Bcl2 antibody (lot No. 56018), and Bax antibody (lot No. 70407) were purchased from Santa Cruz Biotechnology (China). The β-actin antibody (lot No. BS6007M), RAGE antibody (lot No. MB11506), and RBP-3 antibody (lot No. BS8191) were purchased from Bioworld Technology (United States). The V-fluorescein isothiocyanate/propidium iodide (V-FITC/PI) apoptosis kit (lot No. C1062M), terminal deoxynucleotidyl transferase dUTP Nick-end labeling (TUNEL) apoptosis detection Kit (lot No. C1089), 4’,6-Diamidino-2-phenylindole (DAPI) (lot No. C1017), proteinase K (lot No. ST532), and reactive oxygen species (ROS) kit (lot No. S0033S) were purchased from Beyotime Biotechnology (Shanghai, China). Isoflurane (lot No. R510-22) was purchased from RWD Life Sciences (China). Fluorescein Sodium Injection (lot No. 12DLT) was purchased from Shanghai Pioneer Holding Ltd (Shanghai, China).

Preparation of TWF

TWF consists of Astragalus membranaceus (Fisch), AMB, LBL, EBK, CCP, and TAL. The ratio of AMB, LBL, EBK, CCP, and TAL was 6:3:3:2:2. The water-soluble ingredients in TWF were extracted by water decoction method[7]. The main steps were as follows: (1) 20 packages of TWF were immersed in 20 volumes of distilled water for 6 hours; (2) Followed by boiling heating for 3 hours; and (3) This process was repeated 3 times. The boiled solutions were combined, and then freeze-dried after centrifugation (4500 ×g for 30 minutes). Freeze-dried extraction was TWF. The individual sample of AMB, LBL, EBK, CCP, or TAL was prepared according to the method described above.

Determination of ingredients of TWF: The ingredients of TWF were detected using an Agilent 1260 Infinity HPLC with an EliteUQ GIN C18 column (4.6 mm × 250 mm, 5 μm). The mobile phase was acetonitrile (A) and 0.1% formic acid (B). The gradient elution was as follows: (1) 0-5 minutes, 5% A; (2) 5-13 minutes, 5%-15% A; (3) 13-20 minutes, 15%-18% A; (4) 20-23 minutes, 18%-19% A; (5) 23-30 minutes, 19%-20% A; (6) 30-32 minutes, 20%-27% A; (7) 32-37 minutes, 27%-28% A; (8) 37-47 minutes, 28%-35% A; (9) 47-55 minutes, 35%-70% A; and (10) 55-58 minutes, 70%-95% A. The detection wavelength was 203 nm, flow rate was 1.0 mL/minute, and injection volume was 10 µL for each sample. The contents of chemical ingredients were calculated by the reference standards.

Network pharmacology analysis

Eight compounds from TWF and their corresponding targets were screened using TCMSP (https://old.tcmsp-e.com/tcmsp.php), and Swiss Target Prediction (https://www.swisstargetprediction.ch/). The keyword “diabetic retinopathy” was input into the OMIM (https://omim.org/), and GeneCards (https://www.genecards.org/) for screening for DR-related targets. A Venn diagram of drug-related t and disease-related targets was generated via the online tool (https://bioinformatics.psb.ugent.be/webtools/Venn/). Subsequently, the overlapping targets were constructed for the network diagram of drug-component-target and protein-protein interaction (PPI) network via Cytoscape 3.7.2. and STRING (https://cn.string-db.org/). Finally, the common targets were imported into the DAVID database (https://david.ncif-crf.gov/) for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis.

Molecular docking analysis: Molecular docking was performed between the top ten targets in PPI network and eight compounds in TWF using AutoDock Vina 2.0 software. The three-dimensional structure file of eight compounds and target proteins were downloaded from the PubChem database (https://pubchem.ncbi.nlm.nih.gov) and PDB database (https://www.rcsb.org), respectively. The protein targets were removed the water molecules and ligands using PyMol 2.1.0 software, and added the hydrogen atoms to convert to PDBQT files using AutoDock software. The molecular docking results were visualized using PyMOL software.

Experimental analysis in vivo and in vitro

Animals and treatments: Male C57BL/6J mice (4-6 weeks old, 20 ± 2 g) were purchased from Sibeifu Biotechnology Co., Ltd. (Beijing, China), license number was SCXK(Jing)-2019-0010. All mice were housed in a standard animal room with 24-26 °C, 55%-65% relative humidity, and a 12-hour light/12-hour dark photoperiod, and were given ad libitum access to food and water for one week of acclimation. The Animal Ethics Committee of Changchun University of Chinese Medicine approved the animal experiment, and the animals were treated humanely based on the guidelines outlined in the Institutional Animal Care and Use committee and the ARRIVE guidelines (No. 2024251).

DM mice were induced by STZ following the previous reports[7,29]. Total 40 mice were randomly divided into two groups including control group (n = 10) and DM group (n = 30). STZ was freshly dissolved in citrate buffer (0.1 mol/L, sodium citrate, pH 4.3-4.5). Before STZ (50 mg/kg) were intraperitoneally injected, the mice from DM group were fasted for 12 hours, and administered continuously for 5 days. The mice from control group were intraperitoneally injected with citrate buffer. One week after administrating STZ, the mice from DM group were detected the random blood Glu (RBG). DM mice with RBG ≥ 11.1 mmol/L were successfully modeled, which were randomized into three groups (n = 10) based on body weight and RBG, including model group (STZ), CaD group (STZ + CaD capsules, 0.2 g/kg/day), and TWF group (STZ + TWF, 1.0 g/kg/day). The mice from CaD and TWF groups were given the drug for 18 weeks. The mice from control and model groups were given an equal volume of normal saline. The dosage of TWF was determined via the body surface area normalization method for human-to-animal conversion[30], conitored with findings from previous studies[6,7]. Body weight and RBG levels were measured weekly. At the end of the experiment, all mice were anesthetized with 30 mg/kg pentobarbital sodium after fasting for 12 hours.

Fundus fluorescein angiography and optical coherence tomography: Retinal vascular alterations in mice were detected at week 17 using fundus fluorescein angiography (FFA) and optical coherence tomography (OCT). All mice were anesthetized with isoflurane using an ABS-100 animal anesthesia machine (Shanghai Yuyan Instruments, China) with isoflurane. The mice’s eyes were dilated with 1% tropicamide. Retinal microvascular images were captured within 2-5 minutes after intraperitoneal administration of 0.1 mL fluorescein sodium using a Retinal Imaging System (Phoenix Research Labs, United States). Meanwhile, OCT image was captured using the Retinal Imaging System, equipped with Micron IV, Micron Light Source, and Image-guided OCT2[31]. FFA was used to observe microvascular formation and leakage, and OCT was used to measure retinal thickness, including the ganglion cell layer (GCL), inner nuclear layer (INL), and outer nuclear layer (ONL).

Proinflammatory cytokines and tissue factors: After fasting for 12 hours, blood samples were collected from the ophthalmic vein of anesthetized mice (1% pentobarbital sodium). The blood was centrifuged at 12000 rpm for 10 minutes to obtain serum, and the levels of IL-1β, IL-6, TNF-α, VEGF, and HCY were determined using commercially available enzyme-linked immunosorbent assay kits.

Cell culture and cell viability assay of mRECs: The mRECs were treated with control medium (DMEM containing 10% serum and 5 mmol/L Glu, control group) or high Glu medium (DMEM containing 10% serum and 10 mM, 20 mM, 30 mM, 40 mM, or 50 mmol/L Glu) to induce mRECs injury (model group) for 24 hours and 48 hours. Furthermore, 1 × 10⁴ cells/well were treated with 30 mmol/L Glu and co-incubated with TWF (0 μg/mL, 12.5 μg/mL, 25 μg/mL, 50 μg/mL, 100 μg/mL, or 200 μg/mL) for 24 hours and 48 hours. Total 10 μL CCK-8 was added to each well and incubated for 2 hours at 37 °C. Cell viability was calculated based on the optical density at 450 nm using a multimode reader (Infinite M200 Pro, Tecan, Switzerland). The cells were cultured at 37 °C in 5% CO₂.

Cell apoptosis of mouse retina and mRECs: Fresh retinas were peeled off and flushed with phosphate-buffered saline (PBS), ground to form a cell suspension, and filtered through a 70 μm mesh nylon screen. The cells from retinas were centrifuged at 1000 × g for 10 minutes, and then resuspended in PBS at 1 × 10⁶ cells/mL. The mRECs were added to a 24-well plate, and 30 mmol/L Glu with 25 μg/mL TWF (TWF-L), 50 μg/mL TWF (TWF-M), and 100 μg/mL TWF (TWF-H) was incubated for 24 hours and 48 hours. Then, the cells were collected and resuspended in PBS at 1 × 10⁶ cells/mL. The cells from the retinas and mRECs were labeled using the Annexin V-FITC/PI apoptosis kit for 20 minutes. Finally, 10000 events were collected to calculate the apoptosis rate using a DxFLEX flow cytometer (Beckman Coulter Life Sciences).

Histology examination: Retinas, jejunum, and ileum of mice were fixed with 4% paraformaldehyde (PFA). Paraffin sections (4 μm) were prepared and hematoxylin and eosin (HE) staining was performed. Images were captured using an EVOS M7000 Imaging System (ThermoFisher, United States). The thicknesses of the GCL, INL, and ONL were analyzed using the ViewPoint virtual slide viewing software (PreciPoint, Germany).

Immunofluorescence of mouse retina and mRECs: In vivo, retinas were fixed with 4% PFA, 4 μm paraffin sections underwent dewaxing, hydration, antigen retrieval, and blocking with 5% BSA solution for 1 hour. Then the sample was incubated with antibodies including VEGF and ZO-1 for 2 hours. The section was incubated with secondary antibody at room temperature for 1 hour in the dark. Finally, DAPI was added for 10 minutes, and then images were obtained[32]. In vitro, mRECs were inoculated into a 24-well plate containing 30 mmol/L Glu and TWF (25 μg/mL, 50 μg/mL, and 100 μg/mL) for 48 hours. The cells were fixed with 4% PFA for 30 minutes, permeabilized in 0.1% Triton X-100 for 5 minutes, and then blocked with 5% BSA solution for 30 minutes. Cell samples were incubated with antibodies including VEGF and ZO-1. The remaining steps were the same as those described above. These images were captured using the EVOS M7000 Imaging System, and the fluorescence intensity was analyzed using Image J software.

Terminal deoxynucleotidyl TUNEL of mouse retina and mRECs: TUNEL staining was performed using the TUNEL Apoptosis Detection Kit. A 4 μm paraffin section was dewaxed, proteinase K was added for 30 minutes at 37 °C, and then incubated with TUNEL for 60 minutes at 37 °C in the dark[33]. Finally, DAPI was added for 10 minutes and images were obtained. In vitro, mRECs were inoculated into a 24-well plate containing 30 mmol/L Glu and TWF (0 μg/mL, 25 μg/mL, 50 μg/mL, and 100 μg/mL) for 48 hours. The cells were fixed with 4% PFA for 30 minutes, followed by permeabilization with 0.1% Triton X-100 for 5 minutes. TUNEL assay was then performed at 37 °C for 60 minutes in the dark. DAPI was added for 10 minutes, and images were obtained. These images were captured using the EVOS M7000 Imaging System, and the fluorescence intensity was analyzed using Image J software.

ROS: The mRECs (1 × 10⁴ cells/well) were cultured in a 48-well plate with 30 mmol/L Glu and TWF (0 μg/mL, 25 μg/mL, 50 μg/mL, and 100 μg/mL) for 24 hours and 48 hours. Cells were pre-incubated with 2,7-dichlorofluorescein diacetate (10 μM) for 30 minutes in the dark, then trypsinized, and resuspended in RPMI-1640 medium (1 × 10⁶ cells/mL). A total of 10000 events were collected to detect the ROS level using flow cytometry.

Scratch assay: The mRECs (5 × 10⁵ cells/well) were cultured in a 6-well plate for 6 hours, and then a scratch was generated using a sterile 100 μL plastic pipette tip. The cells were inoculated with 30 mmol/L Glu and TWF (0 μg/mL, 25 μg/mL, 50 μg/mL, and 100 μg/mL) for 48 hours at 37 °C in 5% CO₂. Images of wound migration were captured at 0 hour, 24 hours, and 48 hours using Leica Microsystems CMS GmbH (Wetzlar, Germany).

The 16S rRNA gene sequencing and analysis of feces: Fresh feces of mice were obtained at the 18^th week and stored in a -80 °C refrigerator. The 16S rRNA gene sequencing was detected by Novogene Co., Ltd. (Beijing, China). The detailed steps were performed following our previous reports[7,34].

The metabolic profiles of intestinal microbiota by untargeted metabolites analysis: The metabolic profiles were performed by Novogene (https://cn.novogene.com/). The detailed steps were consulted the published article[7,35].

Western blotting: Total protein from the retina was extracted using lysis buffer. Proteins (35 μg) were loaded and separated by SDS-PAGE and subsequently electrotransferred onto a PVDF membrane. The membrane was incubated with primary antibodies, including anti-β-actin, anti-IL-1β, anti-IL-6, anti-TNF-α, anti-VEGF, anti-RAGE, anti-ZO-1, anti-RBP-3, anti-Bax, and anti-Bcl2, followed by incubation with anti-mouse or anti-rabbit secondary antibody for 2 h. The bands were visualized using a FluorChem HD2 imaging system. The relative protein expression was normalized to that of β-actin.

Statistical analysis

Data were statistically analyzed using GraphPad Prism software (version 8.0). All data were presented as mean ± SD. Pearson’s correlation analysis was used to calculate correlation coefficients. Statistical analysis among multiple groups was performed using one-way analysis of variance with the Dunnett’s test. Data of immunoblots and cell/tissue images are representative of at least 3 experiments. P < 0.05 was considered to indicate statistical significance.

RESULTS

Chemical ingredients of TWF

The ingredients of TWF were detected using HPLC. HPLC fingerprints of nine batches of TWF showed 33 common peaks (Supplementary Figure 1), and the fingerprint similarity of TWF was 0.983-0.999 (Supplementary Table 1). Eight compounds were identified using reference substances including p-coumaric acid, calycosin-7-O-β-D-glucoside, isorhamnetin-3-O-neohespeidoside, ononin, calycosin, cinnamic acid, naringenin, and formononetin (Figure 1A). The p-Coumaric acid was derived from LBL and TAL. Calycosin-7-O-β-D-glucoside was derived from AMB. Isorhamnetin-3-O-neohespeidoside was derived from EBK and TAL. Ononin was derived from AMB and CCP. Calycosin was derived from AMB. Cinnamic acid was derived from CCP. Naringenin was derived from EBK, TAL, and CCP. Formononetin was derived from AMB and CCP. Ingredients with higher content were Calycosin-7-O-β-D-glucoside, cinnamic acid, ononin, and isorhamnetin-3-O-neohespeidoside in TWF, which were 48.10 μg/100 mg, 20.92 μg/100 mg, 20.84 μg/100 mg, and 19.96 μg/100 mg, respectively (Table 1).

Figure 1 Chemical ingredients of Tangwang formula were detected by high-pressure liquid chromatography. A: High-pressure liquid chromatography chromatograms of Tangwang formula; B: Venn diagram of drug targets from 8 compounds in Tangwang formula and disease targets from diabetic retinopathy; C: Network diagram of drug-composition-target; D: Protein-protein interaction network diagram of 278 cross targets; E: Degree values of top 10 core targets; F: Top 30 Gene Ontology enrichment analysis of 278 cross targets; G: Top 20 Kyoto Encyclopedia of Genes and Genomes enrichment analysis of 278 cross targets. AGE-RAGE: Advanced glycation end product-receptor advanced glycation end products; AMB: Bunge (Astragali Radix); ATP: Adenosine triphosphate; BP: Biological process; CC: Cellular component; CCP: Cinnamomum cassia Presl (Cinnamomi Ramulus); DR: Diabetic retinopathy; EBK: Eriocaulon buergerianum Koern. (Eriocauli Flos); EGFR: Epidermal growth factor receptor; HIF-1: Hypoxia-inducible factor-1; LBL: Lycium barbarum L. (Lycii Fructus); MAPK: Mitogen-activated protein kinases; MF: Molecular function; RS: Reference substance; TAL: Typha angustifolia L. (Typhae Pollen); TWF: Tangwang formula; VEGF: Vascular endothelial growth factor; 1: P-Coumaric acid; 2: Calycosin-7-O-β-D-glucoside; 3: Isorhamnetin-3-O-neohespeidoside; 4: Ononin; 5: Calycosin; 6: Cinnamic acid; 7: Naringenin; 8: Formononetin.

Table 1 The content of 8 ingredients of Tangwang formula were detected by high-performance liquid chromatography (mean ± SD, n = 3).

Number	Compound name	Standard curve	Content (μg/100 mg)	Source
1	P-Coumaric acid (CAS No. 501-98-4)	y = 21.85x - 5.99, r = 0.99947	17.03 ± 1.34	Lycium barbarum L. (Lycii Fructus), TAL
2	Calycosin-7-O-β-D-glucoside (CAS No. 20633-67-4)	y = 18.38x - 16.94, r = 0.9991	48.10 ± 3.28	AMB
3	Isorhamnetin-3-O-neohespeidoside (CAS No. 55033-90-4)	y = 12.71x + 6.18, r = 0.99983	19.96 ± 0.97	EBK, TAL
4	Ononin (CAS No. 486-62-4)	y = 22.37x - 14.94, r = 0.99987	20.84 ± 1.29	AMB, CCP
5	Calycosin (CAS No. 20575-57-9)	y = 25.05x + 31.40, r = 0.9999	11.71 ± 0.64	AMB
6	Cinnamic acid (CAS No. 140-10-3)	y = 71.62x - 19.12, r = 0.99988	20.92 ± 1.41	CCP
7	Naringenin (CAS No. 480-41-1)	y = 14.49x + 15.74, r = 0.99963	6.48 ± 0.33	EBK, TAL, CCP
8	Formononetin (CAS No. 485-72-3)	y = 32.87x - 37.86, r = 0.99992	4.81 ± 0.42	AMB, CCP

AMB: Bunge (Astragali Radix); CCP: Cinnamomum cassia Presl (Cinnamomi Ramulus); EBK: Eriocaulon buergerianum Koern. (Eriocauli Flos); TAL: Typha angustifolia L. (Typhae Pollen).

Results of network pharmacology analysis

There were 400 drug targets from eight compounds of TWF and 6757 disease targets from DR, and 278 cross targets were obtained between TWF and DR (Figure 1B). Figure 1C showed the network diagram of drug-composition-target between 8 compounds and 278 cross targets. There were 276 nodes and 4519 edges in the PPI network (Figure 1D). The top 10 core targets of TWF treatment DR were AKT1, TP53, ALB, TNF, epidermal growth factor receptor (EGFR), SRC, BCL2, CASP3, STAT3, and ESR1 (Figure 1E). The 278 cross targets were subjected to GO and KEGG enrichment analysis (Figure 1F and G). In GO analysis (Figure 1F), 278 target proteins were mainly enriched in ‘negative regulation of apoptosis process’, ‘insulin receptor signaling pathway’, ‘insulin-like growth factor receptor signaling pathway’, and ‘vascular endothelial growth factor signaling pathway’ in biological process; were located in ‘plasma membrane’, ‘extracellular exosome’, ‘cytosol’ in cellular component; and possessed the function of ‘protein kinase activity’, ‘ATP binding’, ‘enzyme binding’ in molecular function. Top 20 KEGG analysis mainly included ‘AGE-RAGE signaling pathway in diabetic complications’, ‘VEGF signaling pathway’, and ‘apoptosis’ (Figure 1G).

Furthermore, we performed the molecular docking between 8 compounds and 10 core targets to confirm the compound-target interactions. There is a wide agreement that a binding energy less than -5.0 kcal/mol suggests a good binding activity to the receptor, and a binding energy less than -7.0 kcal/mol suggests a strong binding activity. All binding energies of eight compounds docking with 10 core targets were less than -5.0 kcal/mol (Figure 2A), therefore, they had good binding activity. Especially, isorhamnetin-3-O-neohespeidoside, naringenin, ononin, calycosin-7-O-β-D-glucoside, and calycosin showed strong binding with ALB, CASP3, BCL2, EGFR, STAT3, and SCR2. Importantly, BCL2 and CASP3 were the apoptosis-related proteins. Figure 2B and C showed the conformations of 8 compounds with BCL2 and CASP3. Inflammation is the main pathological factor during the occurrence and development of DR. Therefore, TNF was also the important target, Figure 2D showed the conformation of 8 compounds with TNF.

Figure 2 Molecular docking of 8 compounds and 10 core targets. A: Heat map of minimum binding energy (red indicates strong binding activity, blue indicates relatively weak binding activity); B: The conformations of 8 compounds with Bcl2; C: The conformations of 8 compounds with CASP3; D: The conformations of 8 compounds with tumor necrosis factor. EGFR: Epidermal growth factor receptor; TNF: Tumor necrosis factor.

TWF improved the body weight and blood Glu in DR mice

Figure 3A shows the animal experimental protocol. Body weight and RBG level were measured weekly. The body weight of mice in control group maintained a stable increase during 18 weeks (Figure 3B and C), and the RBG level of healthy mice remained within the physiological range (Figure 3D and E). Compared with control group, the body weight of mice in model group was obviously decreased (Figure 3B and C), and the RBG of DR mice showed a significant increase (Figure 3D and E). TWF obviously increased the body weight and reduced the RBG of DR mice (Figure 3B and D); in particular, the body weight and RBG of DR mice in TWF group were significantly improved at the 18^th week (P < 0.05; Figure 3C and E).

Figure 3 Effect of Tangwang formula on the body weight and blood glucose in diabetic retinopathy mice (n = 10). A: Schematic representation of animal study design; B: Changes in body weight per week; C: Statistical analysis of body weight on the 18^th week; D: Changes in random blood glucose per week; E: Statistical analysis of random blood glucose on the 18^th week. The data are expressed as mean ± SD, ^bP < 0.01 vs control group, ^cP < 0.05 vs model group, and ^dP < 0.01 vs model group. CaD: Calcium dobesilate; Glu: Glucose; STZ: Streptozotocin; TWF: Tangwang formula.

TWF inhibited the levels of proinflammatory factors in DR mice and mRECs

High Glu in mice caused an increase in proinflammatory factors, including IL-1β, IL-6, and TNF-α in mouse serum (P < 0.05; Figure 4A) and retina (P < 0.01; Figure 4B). Compared with model group, TWF reduced the levels of IL-1β, IL-6, and TNF-α in the serum (P < 0.05; Figure 4A) and retina (P < 0.01; Figure 4B). These results showed that TWF might inhibit the release of proinflammatory factors in DR mice.

Figure 4 Effect of Tangwang formula on the proinflammatory factors in diabetic retinopathy mice (n = 10) and mouse retinal endothelial cells (n = 3). A: The levels of proinflammatory factors in mouse serum; B: The protein expression of proinflammatory factors in the retina; C: Mouse retinal endothelial cells were cultured in a 96-well plate treated with glucose and Tangwang formula at 24 hours and 48 hours; D: The protein expression of proinflammatory factors in mouse retinal endothelial cells at 48 hours. The data are expressed as mean ± SD, ^aP < 0.05 vs control group, ^bP < 0.01 vs control group, ^cP < 0.05 vs model group, and ^dP < 0.01 vs model group. CaD: Calcium dobesilate; IL: Interleukin; TNF-α: Tumor necrosis factor-α: TWF: Tangwang formula; TWF-L: 30 mmol/L glucose with 25 μg/mL Tangwang formula; TWF-M: 30 mmol/L glucose with 50 μg/mL Tangwang formula; TWF-H: 30 mmol/L glucose with 100 μg/mL Tangwang formula.

In vitro, high Glu was used to culture mRECs, and 12.5 μg/mL, 25 μg/mL, 50 μg/mL, and 100 μg/mL TWF were used to evaluate the viability of mRECs injured by high Glu at 24 hours and 48 hours (Figure 4C). The cell viability was decreased in a time-dose-dependent and Glu-dose-dependent manner, which was 88.78% and 66.67% after treatment with 30 mmol/L Glu at 24 hours and 48 hours, respectively. Therefore, 30 mmol/L Glu was used to establish mRECs injury induced by high Glu. In the TWF toxicity test, TWF at 12.5-200 μg/mL showed no cytotoxicity to mRECs at 24 hours and 48 hours. Notably, 50 μg/mL and 100 μg/mL TWF increased the viability of mRECs injured by 30 mmol/L Glu at 24 hours and 48 hours (P < 0.05), and 25 μg/mL TWF also increased mRECs viability at 48 hours (P < 0.01).

Furthermore, the levels of proinflammatory factors in the mRECs were determined. Figure 4D showed that TWF significantly inhibited the expressions of IL-1β, IL-6, and TNF-α in a dose-dependent manner. Especially, the expression levels of IL-1β, IL-6, and TNF-α in TWF-M and TWF-H groups were lower than those in model group (P < 0.01; Figure 4D). The above results suggested that TWF might attenuate the cell injury and proinflammatory factor expressions of mRECs induced by high Glu.

TWF reduced the retinal injury in DR mice and in mRECs

High Glu caused abnormal expression of some factors and proteins in mouse retinal tissue, particularly VEGF, HCY, RAGE, ZO-1, and RBP-3, which participate in the occurrence and development of diabetic microvascular complications[36,37]. Serum VEGF, serum HCY, retina VEGF, and retina RAGE levels in model group were higher than those in control group (P < 0.01; Figure 5A and B), and ZO-1 and RBP-3 levels in the retina were low expression in model group (P < 0.01; Figure 5B). In the drug treatment group, TWF significantly improved the levels of these factors in DR mice. Compared to model group, TWF reduced the levels of VEGF and HCY in the serum (P < 0.05; Figure 5A), inhibited the production of VEGF and RAGE in the retina (P < 0.01; Figure 5B), and increased the expressions of ZO-1 and RBP-3 (P < 0.05; Figure 5B). Furthermore, immunofluorescence staining was used to verify these results. TWF could inhibit the expression of VEGF and promote the expression of ZO-1 in the retina and mRECs (P < 0.05; Figure 5C and D). These results indicate that TWF could exert a beneficial effect against high Glu-induced retinal damage in DR mice.

Figure 5 Effect of Tangwang formula on retinal injury in diabetic retinopathy mice (n = 10) and mouse retinal endothelial cells (n = 3). A: The levels of tissue factors in mouse serum; B: The protein expression of tissue factors in the retina; C: Immunofluorescence staining of vascular endothelial growth factor and zonula occludens-1 in the retina of diabetic retinopathy mice; D: Immunofluorescence staining of vascular endothelial growth factor and zonula occludens-1 in mouse retinal endothelial cells at 48 hours. ^bP < 0.01 vs control group, ^cP < 0.05 vs model group, and ^dP < 0.01 vs model group. CaD: Calcium dobesilate; HCY: Homocysteine; RAGE: Receptor advanced glycation end products; TWF: Tangwang formula; TWF-L: 30 mmol/L glucose with 25 μg/mL Tangwang formula; TWF-M: 30 mmol/L glucose with 50 μg/mL Tangwang formula; TWF-H: 30 mmol/L glucose with 100 μg/mL Tangwang formula; VEGF: Vascular endothelial growth factor; ZO-1: Zonula occludens-1.

TWF inhibited the cell apoptosis in DR mice and in mRECs

Apoptosis in the mouse retina and mRECs was detected using the Annexin V-FITC/PI kit, TUNEL staining, and western blotting. Our results showed that apoptosis in the retinal tissue of mice induced by STZ (Figure 6A-C) and in mRECs induced by high Glu (Figure 6D-F) was higher than that in control group (P < 0.05). Figure 6A showed that TWF significantly inhibited late apoptosis and total apoptosis of retinal cells compared to model group (P < 0.01; Figure 6A), and the early apoptosis rate in TWF group was also lower; however, there was no significant difference between TWF and model groups (Figure 6A). The positive rate of model mouse retina treated with TWF in the TUNEL assay was also lower than that of the DR mice (P < 0.01; Figure 6B). Moreover, the expression levels of apoptosis-related proteins were measured. TWF significantly reduced the expression of Bax and increased the expression of Bcl2 (P < 0.01; Figure 6C). These results suggested that TWF might inhibit apoptosis of retinal tissue in DR mice.

Figure 6 Effect of Tangwang formula on the cell apoptosis in diabetic retinopathy mice (n = 10) and mouse retinal endothelial cells (n = 3). A: The cell apoptosis of the mouse retina; B: TUNEL staining of mouse retina; C: The protein expression of apoptosis-related proteins Bax and Bcl2 in mouse retina; D: The cell apoptosis of mouse retinal endothelial cells (mRECs) induced by 30 mmol/L glucose at 24 hours and 48 hours; E: TUNEL staining of mRECs at 48 hours; F: The protein expression of the apoptosis-related proteins Bax and Bcl2 in mRECs at 48 hours; G: The reactive oxygen species level of mRECs induced by 30 mmol/L glucose at 24 hours and 48 hours. The red dots indicated the TUNEL-positive apoptosis cell. The data are expressed as mean ± SD, ^aP < 0.05 vs control group, ^bP < 0.01 vs control group, ^cP < 0.05 vs model group, and ^dP < 0.01 vs model group. CaD: Calcium dobesilate; TWF: Tangwang formula; TWF-L: 30 mmol/L glucose with 25 μg/mL Tangwang formula; TWF-M: 30 mmol/L glucose with 50 μg/mL Tangwang formula; TWF-H: 30 mmol/L glucose with 100 μg/mL Tangwang formula.

Furthermore, we verified the anti-apoptotic effects of TWF in mRECs. Three doses of TWF significantly inhibited cell apoptosis at 24 hours and 48 hours, including early, late, and total apoptosis (P < 0.05; Figure 6D). TWF reduced the fluorescence intensity of the mRECs in a dose-dependent manner after 48 hours (P < 0.01; Figure 6E). Meanwhile, TWF-L, TWF-M, and TWF-H inhibited the Bax expression, and promoted the Bcl2 expression in mRECs at 48 hours (P < 0.01; Figure 6F). In addition, we measured the ROS levels of TWF in the mRECs. Figure 6G showed that three doses of TWF reduced the ROS levels of mRECs induced by high Glu at 24 hours and 48 hours (P < 0.05). Therefore, TWF might reduce cell death caused by high Glu by decreasing the apoptosis rate.

TWF improved the retinal microstructure in DR mice

High Glu caused morphological changes and microvascular generation in mouse retina[38]. The results of OCT, FFA, and HE showed that microvascular generation increased (Figure 7A), and the thickness of the retina decreased in model group (P < 0.01; Figure 7B and C). TWF significantly suppressed the formation of retinal microvascular (Figure 7A). Meanwhile, TWF obviously inhibited the thinning of the ONL, INL, and GCL in DR mice induced by STZ (P < 0.05; Figure 7B and C). Cell migration was analyzed using a scratch test. Figure 7D showed that TWF-M and TWF-H significantly promoted the cell migration at 24 hours and 48 hours (P < 0.01; Figure 7D). These results suggested that TWF could improve the retinal microstructure of DR mice and alleviate the retinal injury induced by high Glu.

Figure 7 Effect of Tangwang formula on retinal microstructure in diabetic retinopathy mice (n = 10) and mouse retinal endothelial cells (n = 3). A: Retinal microstructure was detected by optical coherence tomography; B: Retinal microvascular was detected by fundus fluorescein angiography; C: Hematoxylin and eosin staining of retina tissue; D: The images of horizontal migration by scratch assay at 24 and 48 h. The data are expressed as mean ± SD, ^bP < 0.01 vs control group, ^cP < 0.05 vs model group, and ^dP < 0.01 vs model group. CaD: Calcium dobesilate; GCL: Ganglion cell layer; INL: Inner nuclear layer; ONL: Outer nuclear layer; TWF: Tangwang formula; TWF-L: 30 mmol/L glucose with 25 μg/mL Tangwang formula; TWF-M: 30 mmol/L glucose with 50 μg/mL Tangwang formula; TWF-H: 30 mmol/L glucose with 100 μg/mL Tangwang formula.

TWF recovered the relative abundance of abnormal intestinal bacteria in DR mice

A Venn diagram showed that 359, 316, and 323 ASVs were unique in Control, Model, and TWF groups, respectively (Figure 8A). In the alpha-diversity analysis, the indexes of Chao 1, Observed_otus, Simpson, and Shannon in model group showed a decreasing tendency, in particular, the Observed_otus index decreased significantly compared to control group (Figure 8B). A decrease in alpha diversity indexes indicated a decrease in both the richness and evenness of intestinal bacteria in DR mice. TWF could increase the four indexes; Chao 1 and Observed_otus indexes in TWF group had obvious differences compared to model group. Figures 8C-E showed the relative abundance of the main intestinal bacteria in the phylum (top 10), family (top 20), and genus (top 20) among the three groups. Overall, the relative quantity of the main intestinal bacteria in model group showed obvious abnormalities compared to control group; however, TWF could recover the relative quantity of abnormal intestinal bacteria at the phylum, family, and genus levels (Figure 8C-E). TWF significantly decreased the abundance of bacteroidota and increased the level of firmicutes at the phylum level (P < 0.01; Figure 8F). TWF significantly increased the relative abundance of lachnospiraceae, ruminococcaceae and christensenellaceae at the family level (P < 0.05; Figure 8G). Meanwhile, TWF reduced the relative quantity of alcaligenaceae and muribaculaceae at the family level compared to model group (P < 0.05; Figure 8G). At the genus level, TWF significantly elevated the relative abundance of colidextribacter, lachnospiraceae_NK4A136_group, and christensenellaceae_R-7_group, and significantly reduced the relative abundance of oligella (P < 0.05; Figure 8H). TWF also improved the relative abundance of UCG-005, peptococcus, anaerotruncus, oscillibacter, family_XIII_UCG-001, and facklamia in the model mice; however, there were no statistical differences between TWF and model groups (Figure 8H). Specifically, TWF unusually increased the relative abundance of odoribacter at the genus level (Figure 8H). Generally, TWF could restore the relative abundance of abnormal intestinal bacteria in DR mice.

Figure 8 Effect of Tangwang formula on the composition of intestinal bacteria (n = 6). A: Venn diagram of overlap ASV among groups; B: Alpha-diversity of intestinal bacteria; C: The relative abundances of the top 12 bacteria at the phylum level; D: The relative abundances of the top 20 bacteria at the family level; E: The relative abundances of the top 20 bacteria at the genus level; F: The relative abundances of p_bacteroidota and p_firmicutes at the phylum level; G: The relative abundances of 5 bacteria at the family level; H: The relative abundances of 11 bacteria at the genus level (p-phylum, c-class, o-order, f-family, g-genus). The data are expressed as mean ± SD, ^aP < 0.05 vs control group, ^bP < 0.01 vs control group, ^cP < 0.05 vs model group, and ^dP < 0.01 vs model group. TWF: Tangwang formula.

Differential intestinal bacteria were identified using LEfSe analysis (Figure 9). Linear discriminative analysis analysis revealed 12 discriminative features among Control, Model, and TWF groups (linear discriminative analysis score > 4, P < 0.05; Figure 9A). The p_Firmicutes was the key bacterium in control group; p_bacteroidota, c_bacteroidia, o_bacteroidales, f_prevotellaceae, g_alloprevotella were the key bacteria in model group; and c_clostridia, o_lachnospirales, f_lachnospiraceae, f_bacteroidaceae, and g_prevotellaceae_UCG_001 were the key bacteria in TWF group. Figure 9B showed the differential taxa among control, model, and TWF groups.

Figure 9 The LEfSe difference analysis of the relative abundance of intestinal microflora (n = 6). A: Linear discriminative analysis scores; B: Taxonomic cladogram obtained from LEfSe sequence analysis. Biomarker taxa are highlighted by colored circles and shaded areas. Each circle reflects the abundance of taxa in the community (linear discriminative analysis Score > 4, Kruskal-Wallis test, false discovery rate, P < 0.05, p-phylum, c-class, o-order, f-family, g-genus). LDA: Linear discriminative analysis; TWF: Tangwang formula.

TWF improved the metabolic profiles of intestinal microbiota in DR mice

In total, 572 and 1026 peaks were identified in the negative (ES-) and positive (ES+) ion modes, respectively. All metabolites were separated between Model and TWF groups using partial least squares discriminant analysis (partial least squares discriminant analysis, Figure 10A). There were 69 (including 30 up-regulated and 39 down-regulated) and 106 (including 59 up-regulated and 47 down-regulated) differential metabolites between TWF and model groups in the ES- and ES+ ion modes, respectively (Figure 10B). Obviously, TWF could reverse the expression of some of the differential metabolites in DR mice (Figure 10C). The top 20 differential metabolites were shown in Figure 10D. In particular, hypotaurine, 4-chloro-1H-indazol-3-amine, 2,3-dinor prostaglandin E1, 3-benzyl-1-butyl-4-hydroxy-1,2-dihydroquinolin-2-one, and pinocembrin were important up-regulated metabolites, while FAHFA 3:0/18:0, FAHFA 3:0/20:0, NAOrn15:0/16:0, and ciprostene were important down-regulated metabolites (Figure 10D). These differential metabolites were mainly involved in taurine and hypotaurine metabolism, steroid hormone biosynthesis, bile secretion, purine metabolism, and arginine and proline metabolism (Figure 10E). These data indicated that TWF might modulate the fecal metabolites associated with the intestinal microbiota in DR mice.

Figure 10  Tangwang formula changed the metabolic profiles of intestinal bacteria (n = 5). A: Plots of partial least squares discriminant analysis for negative (ES-) and positive (ES+) ions in the model and Tangwang formula (TWF) groups; B: Volcano plot of differentially expressed metabolites between TWF and model groups for ES- and ES+ ions. Red and blue circles indicate significantly up-regulated and down-regulated metabolites, respectively (|fold change| > 1.5, P < 0.05); C: Heatmap of 70 (ES- ion) and 106 (ES+ ion) metabolites among the three groups. The normalized abundance values are depicted visually, where red represents the up-regulated metabolites and blue represents the down-regulated metabolites, respectively; D: Differential metabolites (top 20) in TWF group compared with model group in the ES- and ES+ ions; E: Enriched Kyoto Encyclopedia of Genes and Genomes pathways of differential metabolites in ES- ion and ES+ ion. TWF: Tangwang formula.

Of the 176 differential metabolites, 104 metabolites were involved in 56 categories, including ‘amino acids, peptides, and analogues (amino acids)’, ‘eicosanoids’, ‘fatty acids and conjugates (fatty acids)’, ‘glycerophosphates’, ‘purine deoxyribonucleotides (nucleotides)’, ‘pyrimidine nucleotides (nucleotides)’, ‘bile acids, alcohols and derivatives (bile acids)’, ‘purines and purine derivatives’, and so on.

Data after log2 normalization were used to analyze the correlations between 44 metabolites and 11 bacteria genera using Spearman’s correlation analysis. 44 differential metabolites included 18 amino acids, 5 fatty acids, 6 glycerophospholipids, 3 bile acids, and 12 nucleosides (Figure 11A). The results showed that gamma-Glu-Leu, homoarginine, L-cysteinesulfinic acid, SDMA, gamma-glutamylmethionine, and glycocholic acid were positively correlated with six bacteria genera (anaerotruncus, christensenellaceae_R-7_group, colidextribacter, family_XIII_UCG-001, odoribacter, and UCG-005), which were negatively correlated with oligella. Meanwhile, the 5-fluoro AB-PINACA N-(4-hydroxypentyl) metabolite, erucic acid, heptadecanoic acid, N-tetradecanamide, and LPA 18:0 were negatively correlated with the six bacteria genera above, and positively correlated with oligella. However, there was a weak correlation between the 11 bacteria genera and 12 nucleotides (purine deoxyribonucleotides and pyrimidine nucleotides). These results indicated that anaerotruncus, christensenellaceae_R-7_group, colidextribacter, family_XIII_UCG-001, odoribacter, UCG-005, and oligella were important bacteria genera, which might be involved in regulating the metabolic profiles of 6 amino acids [gamma-Glu-Leu, homoarginine, L-cysteinesulfinic acid, SDMA, gamma-glutamylmethionine, and 5-fluoro AB-PINACA N-(4-hydroxypentyl) metabolite], 3 fatty acids (erucic acid, heptadecanoic acid, and N-tetradecanamide), 1 glycerophospholipids (LPA 18:0), and 1 bile acids (glycocholic acid) in the intestinal microbiota of DR mice. Further, important bacteria genera, metabolites, serum indicators, and apoptosis-related proteins were used to analyze the correlations (Figure 11B). The results showed that anaerotruncus, christensenellaceae_R-7_group, colidextribacter, odoribacter, oligella, L-cysteinesulfinic acid, and glycocholic acid had significant correlations with blood Glu (b-Glu), blood IL-1β, Bax protein and Bcl2 protein. Therefore, intestinal microbiota and metabolic profiles could influence the b-Glu, inflammatory factors, and apoptosis in retina.

Figure 11  Spearman’s correlation analysis. A: Spearman’s correlation analysis between intestinal bacteria and metabolites; B: Spearman’s correlation analysis between intestinal bacteria/metabolites and serum indicators/apoptosis-related proteins. b-glu: Blood glucose; b-IL-1β: Blood interleukin-1β; b-TNF-α: Blood tumor necrosis factor-α; b-HCY: Blood homocysteine; b-VEGF: Blood vascular endothelial growth factor; p-Bax: Bax protein; p-Bcl2: Bcl2 protein in retina.

TWF alleviated the intestinal damage in DR mice

Long-term high Glu environment damages the intestinal mucosal barrier, further lead to jejunum and ileum injury. Our results showed that the crypt depths of jejunum and ileum in model group were higher, meanwhile, villus height, V/C ratio, and goblet cell number were lower than those in control group (P < 0.01; Figure 12). Importantly, TWF obviously increased villus heights and V/C ratios of the jejunum and ileum, and reduced the crypt depths of the jejunum and ileum (P < 0.01; Figure 12A and B). Meanwhile, TWF significantly elevated the number of goblet cells in the jejunum and ileum compared with model group (P < 0.01; Figure 12C and D). The above results suggested that TWF could alleviate intestinal damage.

Figure 12  Tangwang formula improved jejunum and ileum histopathology in diabetic retinopathy mice (n = 6, 40 ×). A: Hematoxylin and eosin staining of the jejunum, and statistical analysis of villus height, crypt depth, and villus height/crypt depth ratio; B: Hematoxylin and eosin staining of the ileum, and statistical analysis of villus height, crypt depth, and villus height/crypt depth ratio; C: Periodic acid-Schiff staining of the jejunum, and the number of goblet cell; D: Periodic acid-Schiff staining of the ileum, and the number of goblet cell. The data are expressed as mean ± SD, ^bP < 0.01 vs control group, and ^dP < 0.01 vs model group. CaD: Calcium dobesilate; TWF: Tangwang formula.

DISCUSSION

STZ, a β-cell toxin, selectively destroys pancreatic insulin-secreting cells, and rapidly induces insulin-deficient diabetes in animals. STZ-induced DR models are pivotal for elucidating the complex pathophysiology of DR. This long hyperglycemic state mimics metabolic dysregulation in DM by triggering a cascade of molecular and cellular events characteristic of DR. In this study, network pharmacology, animal and cell experiments were used to study the ameliorative effects of TWF on DR mice (Figure 12). Our results showed that eight compounds of TWF were identified using HPLC, including p-coumaric acid, calycosin-7-O-β-D-glucoside, isorhamnetin-3-O-neohespeidoside, ononin, calycosin, cinnamic acid, naringenin, and formononetin, which were important active ingredients possibly involved in improving retinal damage. The results of network pharmacology and molecular docking showed that 8 compounds had good binding activity with top 10 core targets, which participated in regulating the multiple signaling pathways related to DR. CASP3 and BCL2, as important core targets, showed strong binding with 8 compounds in TWF. The results of animal experiments showed that TWF increased the body weight, reduced RBG levels of DR mice induced by STZ, and improved retinal injury by reducing the proinflammatory factor levels of IL-1β, IL-6, and TNF-α in the serum and retina, diminishing the expression of VEGF and RAGE, and increasing the expression of ZO-1 and RBP-3. Importantly, we found that TWF significantly reduced the apoptosis rate in the mouse retina, including early apoptosis, late apoptosis, and total apoptosis; meanwhile, TWF decreased the expression of the pro-apoptosis protein Bax, and increased the expression of the anti-apoptotic protein BLC2. Furthermore, the improvement effect of TWF on the retinal microstructure of DR mice was verified by OCT, FFA, and HE staining. These results were validated in mREC induced by high Glu. TWF might inhibit mREC death caused by high Glu, which reduced the cell apoptosis, and release of proinflammatory factors and ROS. Moreover, TWF significantly improved the relative abundance of abnormal intestinal bacteria, and metabolic profiles of the intestinal microbiota in DR mice. Importantly, anaerotruncus, christensenellaceae_R-7_group, colidextribacter, family_XIII_UCG-001, odoribacter, UCG-005, and oligella were involved in regulating the metabolic profiles of bile acids, glycerophosphocholines, amino acids, and fatty acids in the intestinal microbiota of DR mice. These results suggested that TWF alleviated the retinopathy in diabetic mice, potentially by inhibiting inflammation and cell apoptosis, as well as ameliorating dysbiosis of the intestinal microbiota and metabolites.

TWF comprises five Chinese medicines: (1) AMB; (2) LBL; (3) EBK; (4) CCP; and (5) TAL. Jin et al[5] reported that TWF effectively improved microangiopathy and delayed the progression of 192 patients with type 2 diabetes of nonproliferative DR. Eight compounds were identified: The p-coumaric acid, cinnamic acid, calycosin-7-O-β-D-glucoside, isorhamnetin-3-O-neohespeidoside, ononin, calycosin, naringenin, and formononetin were important active ingredients. Yu et al[39] reported that p-coumaric acid decreased the levels of proinflammatory cytokines by inhibiting AGE-RAGE signaling, and attenuated oxidative stress and nephropathy in diabetic rats[40]. Cinnamic acid has the effect of anti-inflammatory, anti-dyslipidemic, and anti-diabetic properties that prevent the development of DM[22]. Calycosin-7-O-β-D-glucoside, isorhamnetin-3-O-neohespeidoside, ononin, calycosin, and formononetin belong to flavonoids with strong anti-inflammatory, anti-oxidant, anti-glycation, and anti-fibrotic activities[41-43]. Wang et al[44] reported that formononetin was absorbed into the blood, and it was an important active ingredient of TWF based on network pharmacology and pharmacokinetics. Therefore, these ingredients from TWF have the potential to be used for the treatment of DR.

The ‘drug-component-target’ strategy of network pharmacology can predict the targets and mechanisms of action[45]. Our results showed that a total of 278 cross targets were identified between TWF and DR, which mainly involved in DR-related signaling pathways including ‘AGE-RAGE signaling pathway in diabetic complications’[25,46], ‘VEGF signaling pathway’[47], ‘apoptosis’[48], and so on. The top 10 core targets of TWF therapy for DR were AKT1, TP53, ALB, TNF, EGFR, SRC, BCL2, CASP3, STAT3, and ESR1, which exhibited good binding activity with eight compounds. In particular, CASP3 and BCL2 displayed strong binding activity with isorhamnetin-3-O-neohespeidoside, naringenin, ononin, and calycosin-7-O-β-D-glucoside. Furthermore, animal and cell experiments validated the potential targets of TWF on DR on apoptosis, VEGF, and AGE-RAGE signaling pathways.

In DM patients, hyperglycemia promotes the formation of AGEs, which bind to RAGE in the retinal cells[46]. AGE-RAGE interactions activate multiple signaling pathways[49]. AGE-RAGE activates nuclear factor kappa B signaling pathways to increase the production of TNF-α, IL-1β, and IL-6, these cytokines further promote inflammation, endothelial cell activation, and immune cells recruitment to the retina, and disrupt normal cellular communication[50]. AGE-RAGE signaling can also upregulate VEGF expression to increase neovascularization[51]. Additionally, it disrupts tight junction proteins, such as ZO-1, which further leads to compromise BRB integrity and increases vascular permeability, thus facilitating the development of retinal edema[52]. Our results showed that TWF significantly inhibited the release of proinflammatory factors (TNF-α, IL-1β, and IL-6), and reduced the expressions of RAGE, VEGF, and ZO-1 in the serum and retina of mice. Furthermore, TWF increased cell viability and inhibited TNF-α, IL-1β, and IL-6 levels in mRECs treated with 30 mmol/L Glu. Taken together, these results suggested that TWF might inhibit the expression and secretion of inflammatory cytokines, as well as alleviate the retinal damage induced by high Glu.

High Glu causes electron leakage in the respiratory chain in mitochondria, which produces a large quantity of ROS. ROS can activate apoptosis signaling pathways, and promote apoptosis by upregulating the expression of pro-apoptotic proteins and downregulating the expression of anti-apoptotic proteins[53]. Our results showed that TWF significantly inhibited apoptosis in the retina, increased Bcl2 expression, and decreased Bax expression in the DR mice. Meanwhile, TWF also reduced the production of ROS and apoptosis rate in mREC induced by high Glu. Therefore, TWF may attenuate apoptosis and oxidative damage both in vitro and in vivo.

FFA and OCT imaging techniques are often used together to detect ophthalmic diseases, such as DR, macular degeneration, and retinopathy of prematurity. FFA highlights the retinal vessels and leakage[54], and OCT captures the cross-section to analyze the changes of retinal thickness, including the GCL, INL, and ONL[55]. Our results showed that sustained high Glu caused a reduction in retinal thickness and an increase in microvascular angiogenesis; however, there was no leakage in mice induced by STZ for 18 weeks. TWF significantly inhibited retinal thinning of the ONL, INL, and GCL and the formation of retinal microvascular. The results of HE staining further verified the beneficial effect of TWF on DR. Therefore, TWF may improve the vascular and retinal injury induced by high Glu.

A high Glu environment can directly damage the intestinal mucosal barrier and increase intestinal permeability[56]. The balance of gut microbiota and metabolites regulates the functions of various organs in the body by the gut-organ axis[57]. The gut microbiota produces various metabolites, including immune factors, neurotransmitters, vitamins, fatty acids, amino acids, bile acids, and so on[58]. These metabolites participate in a complex interplay of neural, hormonal and immune signaling mechanisms[59]. However, imbalance of gut microbiota and metabolites affects the occurrence and development of eye diseases via inflammatory, metabolic and oxidative stress mechanisms by the gut-eye axis[60]. Some bacteria and metabolites, such as lipopolysaccharides, enter the bloodstream to trigger a chronic inflammatory response, disrupt the balance of the intestinal microbiota, and reduce the diversity of the intestinal microflora[61]. The proportion of Firmicutes decreased, and that of Bacteroidota increased in patients[62]. Meanwhile, high Glu gives rise to changes in bile acid, amino acid, fatty acid, and nucleoside metabolism, and so on[63]. Bile acids promote the digestion and absorption of dietary lipids, fatty acids, cholesterol by the gut-liver axis[64]. In our results, anaerotruncus, christensenellaceae_R-7_group, colidextribacter, family_XIII_UCG-001, odoribacter, UCG-005, and oligella were important bacteria genera that regulated the metabolic profiles of bile acids, glycerophosphocholines, amino acids, and fatty acids in the intestinal microbiota of DR mice. TWF significantly increased the abundance of beneficial bacteria including anaerotruncus, christensenellaceae_R-7_group, UCG-005, colidextribacter, and lachnospiraceae_NK4A136_group, which participated in lowering b-Glu and inhibiting the inflammatory factors, and improved the cell apoptosis of retina. Therefore, TWF may potentially alleviate DR-related injury via regulating the intestinal bacteria and their metabolic profiles by the gut-eye axis (Figure 13).

Figure 13  Mechanism of Tangwang formula ameliorates diabetic retinopathy via gut-eye axis. AGEs: Advanced glycation end products; DR: Diabetic retinopathy; IL: Interleukin; NF-κB: Nuclear factor kappa B; NOS: Nitric oxide synthase; ROS: Reactive oxygen species; TNF-α: Tumor necrosis factor-α; TWF: Tangwang formula; ZO-1: Zonula occludens-1.

In this study, we obtained the valuable insights into the mechanisms of TWF, but some limitations should be acknowledged in this study. First, the pharmacokinetics of main chemical components from TWF should be studied to confirm the active ingredients with clear efficacy and optimal dosage for DR. Second, the small samples in intestinal microbiota and metabolic profiles may have introduced bias. The results need to be supported in multiple-dose groups and different animal models, and the number of samples should be increased in further research. Third, the validation of microbiota metabolite functions is the key link to clarify the mechanism of microbiota-host interaction, such as antibiotic treatment or fecal microbiota transplantation, will confirm the function and interaction of intestinal microbiota and metabolites in animal models. Therefore, the potential mechanism of TWF in treating DR needs to be clearly elaborated in future research.

CONCLUSION

TWF improves retinopathy in DR mice induced by STZ by inhibiting the inflammatory response and cell apoptosis, and regulating intestinal microbiota and metabolic dysbiosis. These results support the clinical application of TWF in the treatment of DR. Therefore, this study contributes to the prevention and treatment of DR.

ACKNOWLEDGEMENTS

The authors thank the Guang’anmen Hospital at the Chinese Academy of Chinese Medicine for providing the Tangwang Formula.