Introduction

Rates of stroke incidence have been rising throughout the world owing to economic changes, and these events have been affecting younger individuals1. Strokes are classified into cases of ischemic stroke (IS), cerebral hemorrhage stroke, and subarachnoid hemorrhage stroke. Of these, IS is the most common, with its incidence often being linked to the incidence of cardiovascular and cerebrovascular diseases2. As a treatment method for IS, tissue-type plasminogenactivator has a short therapeutic window and cannot provide benefits for patients who miss the optimal therapeutic window or experience sequelae3,4. There is thus a need to devise alternative therapeutic strategies.

Traditional Chinese medicine (TCM) practices have been extensively implemented for more than 2000 years across several Asian nations. Yangyin Tongnao Granules (YYTN) is a TCM prescription composed of six traditional medicines5: Shengdihuang (Rehmannia glutinosa Libosch.), Shihu (Dendrobium officinale Kimura et Migo), Gegen (Pueraria lobata (Willd.) Ohwi), Huangqi (Astragalus membranaceus (Fisch.) Bge.), Chuanxiong (Ligusticum chuanxiong Hort.) and Shuizhi (Hirudo nipponica Whitman). It had already been approved by the National Medical Products Administration of China in 2003 and the phase II/III clinical trials of YYTN have been executed for IS patients in the acute and convalescent phases. The fingerprint of YYTN extract was obtained by high-performance liquid chromatography (HPLC) previously6,7, which ensured the consistency and stability of the experimental formulation quality. Previous clinical research showed that YYTN could reduce patients’ scores of neurological impairment and improve their cognitive function8. Previous animal studies suggested that YYTN primarily exerts its therapeutic effects via stimulating neurogenesis, providing antioxidant benefits, suppressing inflammation, preventing the apoptotic death of neurons, and other mechanisms9. How this TCM Granules protects against IS at the mechanistic level, however, warrants further study.

Network pharmacology approaches through the construction of protein networks and enrichment analysis have emerged as a robust means of exploring the potential pharmacological and mechanistic basis for the effects of TCM preparations, helping to translate TCM practices into evidence-based medical systems. Such network pharmacology studies are widely used to screen for active TCM ingredients, facilitate drug repositioning, and clarify the complex mechanisms through which TCM preparations exert their effects10.

Building upon the pharmacological potential of YYTN in IS treatment, this study integrated a systematic YYTN-component-target-IS network analysis with molecular docking and experimental validation to comprehensively elucidate the underlying protective mechanisms of YYTN against IS-induced pathological consequences.

Network pharmacology

YYTN ingredient and target selection

Chemical compounds found in YYTN were identified with the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP, http://tcmspw.com/tcmsp.php) and Traditional Chinese Medicines Integrated Database (TCMID, http://www.megabionet.org/tcmid/). Active ingredients were retained if they exhibited an oral bioavailability (OB) ≥ 30% and a drug-likeness (DL) ≥ 0.1811,12. These were then explored in TCMID as well as PubChem (https://pubchem.ncbi.nlm.nih.gov/) for canonical SMILES analyses, selected with the five principles of class drugs (Lipinski’s Rule of Five) in SwissADME (http://www.swissadme.ch/), and introduced into SwissTargetPrediction (http://www.swisstargetprediction.ch) to determine the associated target candidates.

IS-related target selection

IS-related target genes were obtained from Genecards (https://www.genecards.org/), Drugbank (https://www.drugbank.ca/), and DisGeNET (https://www.disgenet.org/) using the keywords “cerebral ischemic stroke” and “cerebral infarction”. Targets in Genecards were screened based on relevance levels exceeding the median. Targets from Drugbank were introduced into UniProt (https://www.uniprot.org/) as a means of obtaining standardized gene names. The results from the three databases were then pooled and deduplicated.

Network construction

YYTN- and IS-related target genes were assessed in Venny2.1.0 (http://bioinfogp.cnb.csic.es/tools/venny/) to determine the intersecting overlapping genes. A YYTN-ingredient-target-IS network was then developed in Cytoscape3.7.2 to gain a more intuitive understanding of the relationships between YYTN and IS.

Protein–protein interaction (PPI) analyses

STRING (https://string-db.org/) was employed to assess interactions among genes and common targets associated with YYTN and IS. The established network was visualized in Cytoscape3.7.2 and subjected to topological analysis.

Functional enrichment analyses

Target genes in the PPI network were analyzed using ClusterProfiler13, org.Hs.eg.db, enrichplot, and ggplot2 in R4.2.3. This approach was used to perform GO analyses of molecular function (MF), biological process (BP), and cellular component (CC) terms. KEGG analyses14,15,16 were also used to explore the biological pathways through which YYTN was predicted to function, focusing on signaling pathways enriched for potential targets. P < 0.05 was deemed statistically significant.

Molecular docking

To expand on the network pharmacology predictions, molecular docking analyses implemented using AutoDock14 were conducted for the four drug compound associated with YYTN with the highest degree values and the four targets with the highest degree values in the PPI network. Good binding was represented by a binding energy < − 5 kcal/mol15. Docking and binding patterns were ultimately generated with Pymol.

Experimental materials and methods

YYTN preparation

YYTN is a national innovative drug and has obtained clinical research approval, approval number: 2003L00206. YYTN comprising six Chinese traditional medicines (Rehmannia glutinosa Libosch.; Dendrobium officinale Kimura et Migo; Pueraria lobata (Willd.) Ohwi; astragalus membranaceus (Fisch.) Bge.; Ligusticum chuanxiong Hort.; Hirudo nipponica Whitman) was provided by Shandong Buchang Pharmaceuticals Co.,Ltd and the batch number was 200602. There was 5.5 g in each package, and it was prepared with sterile water when used. The active compounds in YYTN specimen had been identified by high-performance liquid chromatography (HPLC) with an injection volume of 20 μL, and fow velocity was set at 0.8 mL/min in the previous experiment5. This ensured the reliability of the sample.

Animal

A total of 24 Sprague–Dawley rats (3–4 months old, 250 ± 20 g) from Shanghai SLAC Laboratory Animal Co. LTD were housed in a controlled environment (22 ± 1 °C, 50% ± 10% humidity, 12 h light/dark cycle), with animals being acclimatized for 1 week under routine care. The Animal Experiment Center of Zhejiang Chinese Medical University approved all animal studies (IACUC-20220606-05), which were consistent with the criteria of the Chinese Medical Ethics Committee.

Animal modeling

A modified Zea Longa middle cerebral artery occlusion (MCAO) model16 was established as it provides advantages including the absence of craniotomy, a lack of systemic effects, constant ischemia, and the ability to accurately control the duration of ischemia and reperfusion17. Briefly, rats were anesthetized (intramuscular atropine [0.04 mg/kg] and intraperitoneal Shutai 50 [40 mg/kg]) and a midline neck incision was used for exposure of the common carotid artery (CCA), external carotid artery (ECA), and internal carotid artery (ICA), followed by the insertion of a 0.28 mm polylysine-coated nylon monofilament from the CCA into the ICA until reaching the middle cerebral artery at a depth of 18 ± 2 mm. Following ischemic induction for 1 h, this filament was removed. MCAO modeling success was assessed based on the presence of Horner syndrome in the right eye, circling to the left, and inextensible contralateral paws of the forelimb. Based on the comparison of body surface area in rats and humans, the dose of YYTN selected for use in this study was 1.65 g/kg5. The animals were randomized into various groups (n = 6 /group): control, MCAO model, YYTN treatment (1.65 g/kg), and Edaravone treatment (6.3 mg/kg) groups. The animals in the YYTN and Edaravone groups were given their first dose after 24 hours of ischemia-reperfusion, followed by daily gavage of the appropriate compound for 2 weeks.

ELISAs

Rats were fasted for 12 h following the final treatment, deeply anesthetized, and blood samples were obtained via cardiac puncture. After centrifugation (4500 rpm, 12 min, 4 °C), the sera were retained and stored at − 20 °C. Serum HIF-1α and PAI-1 levels were analyzed with commercial ELISA kits (Jiangsu Enzyme Label Biotechnology Co., LTD., Jiangsu, China).

Western immunoblotting

Samples of homogenized and lysed brain tissue from ratsin each group were centrifuged (12,000 rpm, 5 min, 4 °C) and then analyzed via BCA (Solarbio, pc0020). Samples were then separated via 5% SDS-PAGE, transferred to blots, and probed with primary antibodies specific for β-actin (affinity, AF7018), HIF-1α (28b) (SantaCruz, sc-13515), and VEGFA [VG-1] (abcam, ab1316) and secondary antibodies (Anti-rabbit IgG, HRP-linked (CST, 7074). ECL reagent was used for band development, with protein bands then being analyzed with an ECL chemiluminescence instrument (Clinx Science Instruments Co., Ltd., 610020-9Q) and the ChemiCapture software.

Immunohistochemistry (IHC)

Paraffin-embedded brain sections were deparaffinized, treated with citric acid buffer (pH 6.0, MDL, MD911411) for antigen retrieval, and decolorized using PBS (pH 7.4, MDL, MD911705), treated with 3% hydrogen peroxide for blocking endogenous peroxidase activity, followed by blocking with 3% BSA. They were then probed with primary and secondary antibodies (anti-rabbit IgG, HRP-linked antibody (CST, 7074)) were added for reaction, followed by DAB color development (Servicebio, G1211) and hematoxylin staining. Samples were imaged with an upright fluorescence microscope (Leica, DM3000) and analyzed to clarify how YYTN affects HIF-1α and VEGFA.

qPCR

Trizol was used to isolate total RNA from rat brain tissues, followed by the dissolution of this RNA in 40 µL of DEPC water and the measurement of RNA concentrations. These samples were then used for first-strand cDNA synthesis, followed by qPCR analysis using SYBR Green Master Mix in a PCR instrument (Eppendorf, Mastercycler) with the settings: 95 °C for 10 min; 40 cycles of 95 °C for 15 s, 60 °C for 60 s. The 2−ΔΔCT method was used to determine relative expression, with GAPDH as a normalization control. Utilized primers are shown in Table S1.

Data analyses

Results were analyzed and visualized with SPSS 26.0 and GraphPad Prism 8. If data were normally distributed with homogenous variance, they were analyzed via one-way ANOVAs with Tukey’s test. Data are presented as means ± standard deviation (\(\overline{x }\)± s), with P < 0.05 as the significance threshold.

Results

YYTN target protein identification

A total of 27 compounds in YYTN in the TCMSP database met the criteria for inclusion, including 16, 4, and 7 respectively belonging to Huangqi, Gegen, and Chuanxiong. In addition, 445 compounds in YYTN in the TCMID database met the criteria for inclusion, including 70 belonging to Huangqi, 51 belonging to Gegen, 182 belonging to Chuanxiong, 49 belonging to Shengdihuang, 58 belonging to Shihu, and 35 belonging to Shuizhi. After these two datasets were merged and deduplicated, the remaining ingredients were introduced into PubChem for queries of Canonical SMILES. Based on the three established SwissADME criteria of gateway absorption (high), Lipinski (yes), and Veber (yes), 150 compounds were identified (Table S2). Following the removal of duplicates, 856 predicted protein targets of these compounds were identified with the online SwissTarget Prediction tool at a probability threshold ≥ 0.05 (Table S3).

IS-Related target selection

In total, 3276, 687, and 82 IS-related targets were identified from GeneCards, DisGeNET, and Drugbank, of which 2,428 targets were retained after deduplication (Table S4).

Drug-compound-target-disease network

Venny2.1.0 was used to compare these lists of YYTN-related and IS-related target proteins, ultimately yielding a list of 434 candidate IS-related treatment targets of YYTN (Fig. 1).

Fig. 1
Fig. 1
Full size image

Intersection genes of YYTN and IS.

PPI network establishment

The STRING database was used to generate a PPI network (species: Homo sapiens) followed by visualization with Cytoscape3.7.2 in Fig. 2. Topological analyses were conducted to determine the sizing and color of targets based on degree values, while edge thickness corresponded to the thickness of edge combined scores. The resultant network had 434 nodes, 1,604 edges, 7.39 average node degree, P-value of PPI enrichment < 1.0 exp (− 16). The IS-related targets of YYTN included SRC, PIK3R1, STAT3, MAPK3, VEGFA, HIF and so on. PPI network analysis prioritized hub proteins (PIK3R1/PI3K-AKTpathway, MAPK3/MAPK pathway, VEGFA/VEGF pathway, HIF1A/HIF-1 pathway) as mechanistic candidates.

Fig. 2
Fig. 2
Full size image

PPI Network diagram of YYTN and IS.

Functional enrichment analyses

GO and KEGG analyses were next conducted for the targets identified above, revealing 3,123, 147, and 286 GO-BP, CC, and MF terms, respectively. Based on the adjusted statistical significance P-values, the top 10 most significant terms for each category were plotted (Fig. 3).

Fig. 3
Fig. 3
Full size image

GO-BP/CC/MF enrichment analyses of IS-related targets (Top 10). (Gene counts for individual terms are shown on the x-axis. Pathway names are shown on the y-axis. Adjusted P-values are presented with the color scale.)

The three most highly ranked BP terms were: positive regulation of MAPK cascade (GO: 0043410), response to xenobiotic stimulus (GO: 0009410), and response to oxygen levels (GO: 0070482). The response to oxygen levels entry in the bubble plot had a high enrichment factor and low P-value, providing key clues for understanding the molecular mechanism of YYTN therapy for hypoxia related diseases. The three most highly ranked CC terms were: membrane raft (GO: 0045121), membrane microdomain (GO: 0098857), and synaptic membrane (GO: 0097060). This suggested that YYTN may treat IS by altering the localization of signaling proteins in membrane rafts and regulating the expression of synaptic membrane-associated proteins. The three most highly ranked MF terms were: transmembrane receptor protein tyrosine kinase activity (GO: 0004714), transmembrane receptor protein kinase activity (GO: 0019199), and protein tyrosine kinase activity (GO: 0004713). They indicated that YYTN therapy for IS was likely associated with the tyrosine kinase family, such as VEGFA.

A total of 186 KEGG pathways were found, and the top 15 terms showing the highest statistical significance (adjusted P-value < 0.05) were shown in Fig. 4. Among them, the HIF-1 signaling pathway (hsa04066, P-value = 3.61 exp(-26)) was a key regulatory pathway for cellular hypoxia response. Combining GO analysis and KEGG analysis, YYTN treatment for IS may be related to compensatory responses under hypoxic conditions. The HIF-1 signaling pathway map18,19,20 was plotted (Fig. 5), and the red rectangles represent the genes or proteins enriched in this study. The key genes or proteins enriched in this pathway included STAT3, PI3K, HIF-1α and downstream VEGF and PAI-1, which could guide the selection of key proteins and genes in experiments.

Fig. 4
Fig. 4
Full size image

Dot plot of the pathways of IS in KEGG (Top 15). (The enrichment gene ratio in each pathway is shown on x-axis. The name of each pathway is shown on the y-axis. The gene count according with each term are presented with the size of the dot. Adjusted P-values are presented with the color of the scale.)

Fig. 5
Fig. 5
Full size image

The enrichment results included in the HIF-1 signaling pathway map.

Molecular docking analyses

Molecular docking was used to validate the feasibility of network pharmacology predictions at the structural level of molecular interactions. 3D structures of the four drug compounds (3,9-di-O-methylnissolin, chrysotoxine, erianin and vallesiachotamine) were from the PubChem database. The PBD ID of the four targets (SRC, STAT3, PIK3R1 and MAPK3) were 7NG7, 6NJS, 4JPS, 4QTB from the PDB database. After downloading 3D structures for these compounds from the PubChem database, PyMol was used to remove heteroatoms and water molecules. Molecular docking was performed with AutoDock vina20,21,22, using the position of the original ligand as the docking site. Docking scores were used to select optimal conformations. The good binding activity (< − 5 kcal/mol) between the compounds and the target protein were presented in Table 1. The four optimal binding conformations between the compounds and the target protein (including hydrogen bonds and amino acid residues) were shown in Fig. 6.

Table 1 Molecular docking binding energy.
Full size table
Fig. 6
Fig. 6
Full size image

Schematic diagram of the optimal conformation for molecular docking.

Through molecular docking results and the HIF-1 signaling pathway map (Fig. 5), it was found that the binding energy of STAT3 and PI3K was low. However, HIF-1 was a common downstream indicator of these two targets23,24, and the enrichment of GO and KEGG confirmed hypoxia related processes and activation of the HIF-1 pathway. VEGF and PAI-1 were downstream indicators associated with angiogenesis in HIF-1 pathway25,26. Therefore, the experimental validation prioritized investigating the mechanism by which YYTN ameliorates cerebral ischemia–reperfusion injury through HIF-1 pathway modulation to promote angiogenesis.

Analyses of serum HIF-1α and PAI-1 expression

In ELISAs, serum HIF-1α and PAI-1 levels were measured in different groups of rats (Fig. 7). Significantly higher serum HIF-1α and PAI-1 levels were observed in the Model group relative to control rats (P < 0.05 and P < 0.01, respectively), HIF-1α levels in the YYTN and Edaravone groups were markedly reduced relative to the Model group (P < 0.05). PAI-1 levels were also significantly reduced in the YYTN group relative to the Model group (P < 0.05), with a more marked reduction seen in the Edaravone group (P < 0.01).

Fig. 7
Fig. 7
Full size image

Serum HIF-1α and PAI-1 expression (\(\overline{x }\)±s, n = 6). (Meaning of capital letters in abscissa: (A) Control; (B) Model; (C) YYTN; (D) Edaravone).

Analyses of brain HIF-1α and VEGFA protein expression

Next, Western immunoblotting (WB) was used to evaluate HIF-1α and VEGFA levels in the brains of rats in different groups (Fig. 8). HIF-1α and VEGFA levels in the Model group were markedly raised relative to those in the control group (P < 0.05 and P < 0.01, respectively), while the levels of these proteins in the YYTN and Edaravone groups were significantly elevated (P < 0.01). HIF-1α levels in the Edaravone group were markedly raised relative to the YYTN group (P < 0.05), while the VEGFA levels showed no change (P > 0.05).

Fig. 8
Fig. 8
Full size image

Brain HIF-1α and VEGFA protein expression in WB (\(\overline{x }\)±s, n = 3). ((a) Protein expression cut bands of HIF-1α and VEGFA; (b) histogram of HIF-1α and VEGFA) (meaning of capital letters in abscissa: A: Control; B: Model; C: YYTN; D: Edaravone).

IHC staining revealed a significant increase in the average optical density (AOD) of HIF-1α and VEGFA in the Model group compared with the controls (Fig. 9) (P < 0.05). The AOD for HIF-1α and VEGFA in the YYTN group was markedly raised relative to the Model group (P < 0.05), with an even more significant increase in the Edaravone group (P < 0.01).

Fig. 9
Fig. 9
Full size image

Brain HIF-1α and VEGFA protein expression in IHC (\(\overline{x }\)±s, n = 3). ((a) Staining sections of rat brain tissue with HIF-1α under vertical fluorescent microscope at different magnification and histograms of HIF-1α immunoreactive products AOD in brain tissue; (b) staining sections of rat brain tissue with VEGFA under vertical fluorescent microscope at different magnificationand histograms of HIF-1α immunoreactive products AOD in brain tissue.) (Meaning of capital letters in abscissa: A: Control; B: Model; C: YYTN; D: Edaravone;)

Analyses of brain HIF-1α and VEGFA mRNA levels in qPCR

Significant increases in the mRNA levels of HIF-1α and VEGFA expression were evident in the Model group relative to the controls (Fig. 10) (P < 0.05). No differences in HIF-1α expression were observed when comparing the YYTN and Model groups (P > 0.05), while these levels were markedly raised in the Edaravone treatment group (P < 0.05). VEGFA expression decreased markedly relative to the Model group in response to both YYTN and Edaravone treatment (P < 0.05).

Fig. 10
Fig. 10
Full size image

Histogram of brain HIF-1α and VEGFA gene expression in qPCR (\(\overline{x }\)±s, n = 3). (Meaning of capital letters in abscissa: A: Control; B: Model; C: YYTN; D: Edaravone;)

Discussion

IS occurs as a result of the disruption of arterial blood flow in the brain and consequent hypoxia-related ischemic necrosis in the surrounding brain tissue. This article explored the mechanism of YYTN treating IS through the hypoxia signaling pathway—HIF signaling pathway combining network pharmacology and experimental verification. In this study, YYTN, which consists of 6 traditional TCMs, was subjected to network pharmacology analyses, leading to the identification of 150 bioactive compounds within these medicines. The key ingredients in Huangqi include astragaloside, which can reportedly help prevent and treat cerebrovascular disease by protecting nerve cells and preventing apoptotic death27,28. The key ingredients in Chuanxiong include tetramethylpyrazine, which can prevent or treat IS through vascular expansion, neuroprotection, and antioxidant effects29,30. The key ingredients in Shihu include dendrobine, which can lower blood pressure to prevent IS31,32. Daidzein, derived from Gegen, can contribute to improved neurological function scores and to significant reductions in blood fibrinogen levels in cerebral infarction patients30,33.

GO enrichment analysis-derived BP terms highlighted the important role of MAPK signaling in stroke, suggesting that drugs and genes associated with this pathway may be relevant to the alleviation of ischemic brain injury34. With respect to CC terms, membrane lipid rafts are very important for survival signaling in neurons35. Among MF terms, protein tyrosine phosphatase 1B-induced microglia activation can lead to the worsening of neuroinflammation and aggravated brain injury36. Based on network pharmacology and KEGG enrichment analyses, various key proteins were identified including HIF-1α, VEGFA, and PAI-1, which are related to the HIF-1 signaling pathway, as has been reported previously. HIF-1 is a key regulator of hypoxic activity, influencing the expression of more than 700 target genes associated with both adaptive and damaging processes37. HIF-1 can regulate both VEGF and PAI-138. Network pharmacology can search for the main active ingredients of traditional Chinese medicine formulas and their mechanisms of action in treating diseases39. However, network pharmacology also has limitations, as its predictions depend on the comprehensiveness and accuracy of the reference database. For example, owing to the overly large chemical formula of hirudin, which is the primary ingredient in Shuizhi, its structure could not be determined such that it was omitted from these analyses. Inaccurate or blank data in these databases may affect the reliability of the results. Therefore, other methods such as experiments are needed to validate these findings40.

After completing the above network pharmacology and bioinformatics analyses, experimental validation was performed through a series of Western blotting, IHC, and qPCR analyses demonstrating that YYTN was able to treat IS through the enhancement of brain HIF expression. Edaravone is a potent antioxidant that has been proposed as an effective component of stroke care in Asian nations41, and it was thus selected here as a positive control drug. The upregulation of HIF-1α can reduce neuronal apoptosis and improve systemic IS resistance42. YYTN-based IS treatment was associated with increased VEGFA protein levels in the brains of MCAO model rats as determined through IHC and Western blotting. Higher levels of VEGFA have been demonstrated to facilitate collateral circulation, contributing to better angiogenesis and post-IS recovery43. Astragaloside IV can reportedly improve VEGF levels and markedly bolster HIF/VEGF pathway activity28. Under YYTN and edaravone treatment in qPCR, however, brain VEGFA levels decreased. Significant reductions in VEGFA expression levels have been reported during cerebral ischemia44. Alternative splicing can also give rise to multiple subtypes of VEGFA45. Additional studies will be necessary to overcome the limitations of the present analyses. ELISAs performed herein revealed that serum PAI-1 levels declined in response to YYTN and edaravone treatment. PAI-1 deficiency can reportedly lead to reductions in infarct size and improved recovery of cerebral blood flow46. However, acute IS patients reportedly exhibit higher PAI-1 levels, and no differences have been reported among cerebral infarction clinical subgroups47. PAI-1 may thus be an effective biomarker of IS incidence but a poor indicator of stroke length or severity.

Conclusion

In conclusion, the network pharmacology analyses performed herein offer new insights into the multi-component multi-target nature of YYTN-mediated IS treatment. Molecular docking results provided further support for these network pharmacology predictions, while animal model experiments confirmed the efficacy of YYTN as an IS treatment capable of improving cerebral arterial blood flow interruptions and protecting against hypoxic-ischemic necrosis in the brain through the regulation of HIF-1α and downstream targets including VEGFA and PAI-1. These insights into the therapeutic effects of YYTN in a model of IS can serve as a foundation for efforts to develop new treatments for IS. In the future, in-depth studies will be performed to validate and expand on these findings at the cellular level.