The Effect of Hedysarum multijugum Maxim.-Chuanxiong rhizoma Compound on Ischemic Stroke: A Research Based on Network and Experimental Pharmacology

Background Hedysarum multijugum Maxim.-Chuanxiong rhizoma compound (HCC) is a common herbal formula modified from Buyang Huanwu decoction. Clinical trials have demonstrated its therapeutic potential for ischemic stroke (IS). However, the mechanism of HCC remains unclear. Methods The HCC's components were collected from the TCMSP database and TCM@Taiwan database. After that, the HCC's compound targets were predicted by PharmMapper. The IS-related genes were obtained from GeneCards, and OMIM and the protein-protein interaction (PPI) data of HCC's targets and IS genes were obtained from the String database. After that, the DAVID platform was applied for Gene Ontology (GO) enrichment analysis and pathway enrichment analysis and the Cytoscape 3.7.2 was utilized to construct and analyze the networks. Finally, a series of animal experiments were carried out to validate the prediction results of network pharmacology. The expressions of GRP78, p-PERK, and CHOP proteins and mRNAs in different time periods after HCC intervention were detected by Western blot, immunohistochemistry, and RT-qPCR. Results A total of 440 potential targets and 388 IS genes were obtained. The results of HCC-IS PPI network analysis showed that HCC may regulate IS-related targets (such as ALB, AKT1, MMP9, IGF1, and CASP3), biological processes (such as endoplasmic reticulum stress, inflammation modules, hypoxia modules, regulation of neuronal apoptosis and proliferation, and angiogenesis), and signaling pathways (such as PI3K-Akt, FoxO, TNF, HIF-1, and Rap1 signaling). The animal experiments showed that HCC can improve the neurobehavioral scores and protect the neurons of IS rats (P < 0.05). HCC inhibited the expression of p-PERK in the PERK pathway from 12 h after surgery, significantly promoted the expression of GRP78 protein, and inhibited the expression of CHOP protein after surgery, especially at 24 h after surgery (P < 0.05). The results of RT-qPCR showed that HCC can significantly reduce the expression of CHOP mRNA in the neurons in the CA1 region of the hippocampus 72 h after MCAO (P < 0.05). Conclusion HCC may achieve a role in the treatment of IS by intervening in a series of targets, signaling pathways, and biological processes such as inflammation, oxidative stress, endoplasmic reticulum stress, and angiogenesis.


Introduction
Globally, stroke is the leading cause of death and disability, and the cost of medical resources after stroke is high [1,2]. Cerebral infarction (CI) accounts for 70% to 80% of all strokes, while hemorrhagic strokes account for less than 30%. CI, also known as ischemic stroke (IS), is an ischemic necrosis of the brain tissue at the corresponding site after cerebral artery occlusion, which can be complicated by postinfarction hemorrhage [3,4]. The pathogenesis of IS is thrombosis or embolism, which causes brain tissue ischemia and hypoxia in this blood supply area, leading to nerve cell death and loss of function [4][5][6]. The symptoms vary depending on the blood vessels involved in the lesion [4]. Current research showed that the main risk factors leading to stroke are diabetes, dyslipidemia, and hypertension, which play an important role in the development of IS [4][5][6]. The current treatment strategies for IS are mainly (1) acute phase: early improvement of blood circulation in the ischemic area and promotion of neurological recovery; (2) recovery period: continue to strengthen the physical exercise and speech function training of the limbs [7]. In addition to drugs, it can be combined with physical therapy and alternative medicine. Since there is no specific drug, the treatment after IS was still in the focus of research and development. The rise of alternative medicine has gradually provided new strategies for post-IS treatment, especially in developing countries, where cheap and available Chinese herbal medicines are an important choice for patients [8]. Traditional Chinese medicine represented by complex preparation has accumulated a large number of clinical medical treatment practices in the treatment of IS and gradually developed important and effective prescriptions [9]. Hedysarum multijugum Maxim.-Chuanxiong rhizoma compound (HCC) was first used by the First Affiliated Hospital of Hunan University of Chinese Medicine. It was modified from the classic prescription Buyang Huanwu decoction and has applied for a patented product, which is mainly composed of Hedysarum multijugum Maxim., Chuanxiong rhizoma, Pheretima, and Bombyx batryticatus [10]. Current studies show that HCC can reduce serum TNF-α and plasma TXB2 levels and increase plasma 6-keto-PGF1α levels in IS patients, suggesting that its therapeutic effect may be related to regulation of vascular endothelial/platelet function balance [11,12]. In an in vitro model, HCC extract was resistant to rhTNF-α-induced apoptosis in cultured HUEC and the effect of rhTNF-α on anticoagulant activity. However, its mechanism for treating IS needs further disclosure [13].
Systematic pharmacological studies have revealed the biological networks in which drugs work. The integration strategy of network biology and multidirectional pharmacology is conducive to expanding the available drug target space and is expected to become one of the new strategies for drug discovery, especially in the field of research and development of Chinese medicine formulae [14][15][16][17][18]. Our previous research successfully analyzed the molecular mechanisms and pharmacodynamic components of traditional Chinese medicine in the fields of cardiovascular and cerebrovascular diseases, tumors, and complex endocrine diseases through the chemical informatics and complex pharmacology methodological strategies [17,18]. Therefore, in this research, we hope to explore the biological mechanism of endoplasmic reticulum stress after HCC intervention in IS by systematic pharmacology. The research process is shown in Figure 1.

Prediction of the Mechanism of HCC Intervention in
Since the use of the ADME model to predict the potential compounds of HCC has limitations [27], in order to avoid the omission of potential compounds, we searched a large number of references and included oral absorbable compounds with pharmacological activity. Combined with relevant references [28,29], several compounds are included.
The mol2 format file of the potential compounds molecular structure was imported into PharmMapper (http://lilab-    5 Oxidative Medicine and Cellular Longevity ecust.cn/pharmmapper/) to predict the potential target of HCC [30]. The PDB ID of the HCC protein target is introduced into the UniProt database (https://www.uniprot.org/ uniprot/) with the species limited to "Homo sapiens" to search for the official symbols of each protein target (Table S1, see Supplementary Materials).

Ischemic Stroke Gene.
To construct the biological network of IS, the IS-related genes were collected from the OMIM database (http://omim.org/) [31] and GeneCards (http://www.genecards.org) [32]. Finally, one thousand and six hundred and thirty-five (1635) IS-related genes were obtained. These IS-related genes with relevance score > 2:5 Figure 4: HCC-IS PPI network (green circle stands for HCC-IS; blue circle stands for IS genes; orange circle stands for HCC targets). 6 Oxidative Medicine and Cellular Longevity were used for subsequent biological network construction and network analysis (Table S2).

Protein-Protein Interaction
Data. The String database (https://string-db.org/, Version 11) is a database containing known and predicted protein-protein interaction (PPI) data [33]. The potential targets of HCC were introduced into the String database with the species limited to "Homo sapiens" and the PPI data of them were obtained. The result is saved in TSV format. The node 1, node 2, and combined score information in the file is retained and imported into Cytoscape software to draw the interaction network [17,18,34].

Network
Construction and Analysis. The Cytoscape software ver. 3.7.0 was used for network construction. In Cytoscape, each node represents a gene, protein, or molecule, and the node-to-node relationship (edges) represents the interaction between these biomolecules. The degree of the node represents the number of nodes connected to the node in the network. The greater the degree is, the more likely this target becomes the key target in the network [34].
In the interaction network, there are several regions where proteins interact closely; such regions are called cluster [34]. These different clusters represent different biological modules, and it is possible to treat diseases by intervening in these modules. The plugin of Cytoscape, MCODE, was utilized to detect clusters in PPI network [17,18]. DAVID (https://david.ncifcrf.gov/summary.jsp, Version 6.8) provides systematic, comprehensive biofunctional annotation information for large-scale genes or proteins to identify the most significant enriched biological annotations. This database was utilized to undergo Gene Ontology (GO) enrichment analysis and pathway enrichment analysis for these networks [35].

Model Preparation.
Modeling was carried according to the improved Zea Longa intraluminal tether block method [36]. The production of the wire plug is as follows: the diameter of the monofilament nylon wire is 0.22-0.24 mm, and each segment is cut into a length of 4-5 cm; disinfect the line with alcohol and soak it in heparin sodium. The rat focal cerebral ischemia model was prepared by the middle cerebral artery occlusion (MCAO) method: (1) anesthesia: adult SD rats were anesthetized by intraperitoneal injection of 4% sodium pentobarbital (50 mg/kg); (2) fixation and disinfection: after complete anesthesia, the rat is placed in a supine position on a rat fixation plate, sheared, and the skin is disinfected with complex iodine; (3) incision: take the midline incision of the neck, generally 2.5-3.5 cm; (4) isolation and ligation of blood vessels: the right sternocleidomastoid muscle was bluntly separated, and the right common carotid artery (CCA) was separated with forceps and the proximal end of CCA was ligated; locate the CCA bifurcation and separate the inwardly communicating external carotid artery (ECA) and the outwardly moving internal carotid artery (ICA), ligature the branch of ECA, and then ligature the ECA; (5) place the line: the ICA was clamped with an arterial clip, and then a small opening was made 4 mm below the CCA bifurcation. The artery clamp was released and the iliac line (0.28 mm diameter fishing line, head end 0.34 mm) was inserted into the ICA with an insertion depth of 22 ± 0:5 mm. Finally, the telecentric end of ICA and CCA is ligated; and (6) stitching: cut off the extra tying line, sprinkle a small amount of penicillin powder in the incision, suture the incision layer by layer, and clean the incision. The blank group does not insert the thread plug, and the remaining steps are the same as the model group. After anesthesia was awoken, the rat's eyelids drooped, showing a significant Horner syndrome. Then TTC staining was used to judge whether the modeling was successful. After the operation, the rats were kept in a single cage, and the breeding room was kept quiet and constant temperature and humidity.
2.5.2. Animal Grouping and Intervention Methods. Two hundred and seventy (270) healthy male SPF SD rats were randomly divided into 6 groups (45 in each group): the blank group (the sham operation group), the model group (the IS group), the HCC low-dose group (4.5 g crude drug/kg), the HCC medium-dose group (9 g crude drug/kg), the HCC high-dose group (18 g crude drug/kg),  Oxidative Medicine and Cellular Longevity and the nimodipine group (32 mg/kg). The MCAO model was prepared for each of the other groups except the blank group. Each group was given the corresponding drug by oral gavage per 100 g body weight once a day 7 days before surgery, and the model group was given normal saline (NS). The drug was administered continuously 7 days before the modeling, and the last time was 1 hour before the modeling. The HCC low-, medium-, and highdose groups were given 4.5 g crude drug/kg, 9 g crude drug/kg, and 18 g crude drug/kg concentration of HCC extract, respectively.
2.5.3. Neurobehavioral Score. Neurobehavioral scores are based on the Zea Longa 4-point scale [36]. Neurobehavioral examination was divided into 5 grades: 0 point: normal, no neurological deficit; 1 point: the left front paw cannot be fully extended, mild neurological deficit; 2 points: when walking, the rat turned to the left side (temporal side), moderate neurological deficit; 3 points: when walking, the rats were dumped to the left side (temporal side), and severe neurological deficits were observed; and 4 points: cannot walk, lose consciousness.
Gray value analysis is as follows: six slices were randomly taken from each animal, and six fields were randomly taken 9 Oxidative Medicine and Cellular Longevity under a 20-fold objective lens. The image was processed by the Motic image analysis system, and the gray value of each field of view was measured and calculated. The average gray value of each group of animals was compared (gray value range 0-255; the larger the gray value is, the lower the expression level of the positive product is).
2.5.6. Western Blot Detection. The rat hippocampus was taken, homogenized by adding 5 volumes of buffer, and centrifuged, and the supernatant was taken. The protein concentration was quantified using the Lowry method protein and then subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The order of the sampling was as follows: the IS group 6 h after surgery (M6), 12 h after surgery (M12), 24 h after surgery (M24), and 48 h after surgery (M48); the HCC group 6 h after surgery (HCC6), 12 h after surgery (HCC12), 24 h after surgery (HCC24), and 48 h after surgery (HCC48).
Immunoassay is described as follows: the target nitrocellulose membrane was placed in a plastic bag of moderate size, and a blocking solution (3% BSA+5% skim milk) was added and shaken slowly for 1 h at room temperature, and was washed 3 times with 0.02 M PBS for 10 min each time. Then, 1 : 500 rabbit anti-GRP78 polyclonal antibody, rabbit anti-p-PERK polyclonal antibody, mouse anti-GADD 153 (CHOP) monoclonal antibody, and 1 : 1000 mouse anti-GAPDH monoclonal antibody 5-8 ml were added in an amount of 0.1 ml/cm 2 according to the membrane area and incubated at 4°C overnight. The membrane was taken out the next day and washed 3 times with 0.01 M PBS for 10 min each time; continue to add horseradish peroxidase-labeled goat anti-rabbit or goat anti-mouse secondary antibody (1 : 1000) according to the membrane area in an amount of 0.1 ml/cm 2 and incubate for 1.5 h at room temperature; after that, the membranes were washed 3 times with 0.02 M PBS for 10 min each time. Subsequently, color  Figure 7: Herb-biological processes-target network (red diamond stands for biological processes; purple circle stands for HCC-IS; blue circle stands for IS genes; pink circle stands for HCC targets. The gray lines stand for the relationship among herbs and targets; the black lines stand for the relationships among biological processes and targets).  11 Oxidative Medicine and Cellular Longevity development, exposure, and photography were carried out using an electrochemical-enhanced luminescence (ECL) immunoassay kit. The results were analyzed by IPP6.0 image analysis system, and the corresponding IOD value and the ratio of glyceraldehyde-3-phosphate dehydrogenase (GADPH) were obtained. Protein expression was positively correlated with IOD values.
2.5.7. RT-qPCR Detection of Related Gene Expression. The total mRNA was extracted from the tissues extracted with TRIzol, and the RNA purity was identified by 1.2% agarose gel electrophoresis. Subsequently, the first-strand cDNA synthesis was carried out: cDNA was synthesized by reverse transcription using the total tissue mRNA as a template, and the obtained cDNA was stored at -20°C. Then, a real-time PCR test is performed. The amplification procedure is described as follows: 94°C for 4 min, 94°C for 40 sec, and 60°C for 30 sec; the CT value was obtained, and the relative expression amount of the target gene was calculated by the 2 -ΔΔCT method. QT-PCR primer sequence is described as follows: the gene sequence was downloaded from the NCBI database. The qPCR primers were designed using Primer 5.0 software. The primer sequences are shown in Table 3.

12
Oxidative Medicine and Cellular Longevity 2.6. Statistical Analysis. All data are expressed as the mean ± SD. Statistical analysis was performed using SPSS 19.0 software. For multiple comparisons, one-way analysis of variance (ANOVA) was performed and the comparison between the two groups was analyzed using the LSD test. P < 0:05 was considered to be statistically significant.

Biological Processes of HCC-IS PPI Network.
The HCC-IS PPI network was analyzed by MCODE to obtain the clusters. The clusters of this network are shown in Figure 5 and Table 4. The genes and targets in these clusters were put into the DAVID database to undergo GO enrichment analysis as an example.
After the GO enrichment analysis, a lot of biological processes of each cluster were returned. Cluster 1 is associated with inflammation modules, hypoxia modules, regulation of neuronal apoptosis and proliferation, angiogenesis, coagulation and platelet activation, oxidative stress, and so on. Cluster 2 is associated with coagulation, inflammation, hypoxia, angiogenesis, endothelial cell proliferation, response to estrogen, and nitric oxide biosynthesis. Cluster 3 is involved in the regulation of coagulation, fibrinolysis, negative regulation of endothelial cell apoptosis, oxidative stress, neuronal production, and regulation of vascular and inflammatory responses. Cluster 4 is related to endoplasmic reticulum stress and neuronal apoptosis. Cluster 5 is related to lipid synthesis and metabolism such as cholesterol and triglycerides. Cluster 8 is involved in the metabolism of lipids. Cluster 10 is related to redox reactions such as glutathione. Cluster 12 is associated with the metabolism of steroid hormones. Cluster 13 is associated with the metabolism of steroid hormones. Cluster 15 is related to lipid metabolism. Cluster 18 is related to endoplasmic reticulum stress. Clusters 6,9,11,14, and 16 did not return any IS-related biological processes. Cluster 7 failed to return any human biological processes (Table S3). Since Cluster 4 contains many classic biological processes, bubble chart is created using the main biological process data contained in Cluster 4 (Figures 6 and 7).
In summary, through the analysis of the HCC-IS PPI network, we reveal the potential mechanism of HCC treatment of IS, which is mainly related to inflammation, hypoxia, endoplasmic reticulum stress, oxidative stress, angiogenesis, coagulation and platelet activation, angiogenesis, endothelial damage, the negative regulation of endothelial cell apoptosis, and the regulation of angiogenesis and inflammatory response after IS. These biological processes occur mainly in the early stages of IS [37][38][39][40]. These directions can be developed in the future when researching IS treatment strategies.

Signaling
Pathways of HCC-IS PPI Network. The HCC targets combining with IS genes were put into the DAVID database for pathway enrichment analysis. After this, thirtytwo (32) IS-related signaling pathways were returned ( Figure 8).
These signaling pathways are ranked according to the degree of enrichment (negative correlation with P value) changes of IS. For example, it is involved in the signaling pathways such as complement and coagulation cascades, and platelet activation that occur in the early stage of IS [41]. The PI3K-Akt signaling pathway is associated with inflammation, endoplasmic reticulum stress, and neuronal apoptosis in neurovascular units [42]. The FoxO signaling pathway is related to endoplasmic reticulum stress and oxidative stress [43]. The TNF signaling pathway and NF-κB signaling pathway are related to inflammation [44]. The HIF-1 signaling pathway is related to angiogenesis [45]. The neurotrophin signaling pathway is associated with neural stem cell differentiation and synaptic regeneration [46]. The PPAR signaling pathway is related to energy metabolism [47]. After IS, neurovascular units (which includes the neurons, blood-brain barrier, microglia, and extracellular matrix components; there is a wide range of signal connections between them, which is the material basis for ensuring neuron function and normal cerebral blood flow) play a very important role in the pathophysiology and clinical treatment of IS [48][49][50]. The neurons, blood-brain barrier, and microglia are intertwined through various cascade response signaling pathways after ischemia and hypoxia to jointly promote the occurrence, development, and outcome of IS [51,52].
In summary, this study explored the mechanism of HCC intervention in IS through network pharmacology strategies; in order to further reveal the specific biological processes and targets related to the therapeutic effect of HCC, we conducted MCODE analysis on the HCC-IS PPI network to obtain clusters of biological modules. After network analysis and literature mining, we found that there are few studies on the role of endoplasmic reticulum stress biological modules in IS ( Figure 10). In this study, endoplasmic reticulum stress is one of the most important biological modules in the HCC-IS PPI network, and its cluster ranks fourth. The biological modules related to the first three clusters (inflammation, angiogenesis, etc.) have been studied a lot, and there are related therapeutic drugs. Therefore, this study will use IS animal model experiments to verify and study the main core targets [GRP78 (HSPA5), p-PERK (EIF2AK3), and CHOP (DDIT3)] in the endoplasmic reticulum stress biological module.
3.3. Neurobehavioral Score. The results showed that the neurobehavioral scores of the model group were significantly higher than those of the sham operation group (P < 0:05). Compared with the model group, each dose of HCC and nimodipine can significantly reduce the neurobehavioral score of rats after MCAO (P < 0:05). Compared with the low-dose group and the middle-dose group, the HCC highdose group was more effective, and the difference was statistically significant (P < 0:05) (Figure 11).

Morphological Observation.
Nissl staining showed that the Nissl body was located in the cytoplasm and was purple-red. Compared with the model group, the HCC high-, medium-, and low-dose groups and the nimodipine group had clear cell boundaries, and the cytoplasmic Nissl bodies were evenly distributed, and the nucleoli were clearly visible. HCC and nimodipine can significantly increase the number of the neurons surviving per unit area of the hippocampal CA1 area, and the difference is statistically significant; and compared with the low-dose group and the middle-dose group, the HCC high-dose group was more effective (P < 0:05) (Figures 12 and 13).
3.5. The Expressions of GRP78, p-PERK, and CHOP Protein Detected by Immunohistochemistry 3.5.1. Expression of GRP78 Protein. The average gray value of each group of animals was compared (gray value range 0-255; the larger the gray value is, the lower the expression level of the positive product is). GRP78-positive products are brown-yellow particles located in the cytoplasm of the neurons. Compared with the sham operation group, the GRP78 in the model group gradually increased the expression level from 6 hours after surgery, reached the peak at 24 hours, and was significantly downregulated after 72 hours (P < 0:05). After HCC intervention, the expression of GRP78 was significantly upregulated after operation, and the expression level was still higher than that of the model group and nimodipine group at 72 hours (P < 0:05) (Figures 14-16).

Expression of p-PERK Protein.
The p-PERK protein is brownish yellow granules, mainly expressed in the cytoplasm, and has almost no expression in the sham operation group. Compared with the sham operation group, the expression of MCAO in the model group increased at 6 hours after operation and then began to decrease; the difference was statistically significant (P < 0:05). After HCC intervention, the expression of p-PERK protein was significantly lower than that of the model group at 6 hours after surgery and then gradually increased; it was significantly higher than the model group and the nimodipine group after 12 hours and maintained at a high level (P < 0:05) (Figures 17-19).

Expression of CHOP Protein.
The CHOP-positive product is a brownish granule located in the nucleus of the neuron. There was almost no expression in the sham operation group. Compared with the sham operation group, the expression level of the model group increased gradually from 6 hours after MCAO, and the difference was statistically significant (P < 0:05). Compared with the model group, the expression of CHOP after HCC intervention was significantly downregulated from 12 hours after MCAO, the difference was statistically significant (P < 0:05), and the expression of CHOP was significantly lower at 12 hours after the HCC intervention than in the nimodipine group (P < 0:05). (Figures 20-22).
3.6. The Expressions of GRP78, p-PERK, and CHOP Protein Detected by Western Blot. Compared with the sham operation group, the expression of GRP78 protein in the model group was significantly upregulated, and the difference was statistically significant (P < 0:05). Compared with the model group, the expression of GRP78 protein was significantly upregulated at each time after MCAO intervention, and the difference was statistically significant (P < 0:05). Compared with the sham operation group, the expression of p-PERK its downstream proteins eIF2α and CHOP, and then inhibit the apoptosis of the basal forebrain neurons in rats after MCAO [59]. In this study, the expression of CHOP in the neurons caused by PRK activation can also be downregulated by HCC. However, the time of inhibition of CHOP protein and genes is not the same, which requires further

21
Oxidative Medicine and Cellular Longevity experiments to explore its mechanism. Cao et al. found that Yiqi Fumai (YQFM) powder, an extract of Shengmai powder, can inhibit the expression of the ERS pathway marker protein CHOP [60]. Our research also shows that HCC has a certain effect on improving the behavior of rats after IS, and it can significantly increase the survival of the hippocampal CA1 neurons after MCAO. HCC can inhibit the expression of p-PERK in the PERK pathway, significantly promote the expression of GRP78 protein, and inhibit the expression of CHOP protein postoperatively, especially at 24 h postoperatively. The results of RT-qPCR also showed that HCC can significantly reduce the expression of CHOP in the hippocampal CA1 neurons 72 h after MCAO.

Conclusions
HCC may achieve a role in the treatment of IS by intervening in a series of targets (such as ALB, AKT1, MMP9, IGF1, and CASP3), biological processes (such as endoplasmic reticulum stress, inflammation modules, hypoxia modules, regulation of neuronal apoptosis and proliferation, angiogenesis, coagulation and platelet activation, and oxidative stress), and signaling pathways (such as PI3K-Akt, FoxO, TNF, HIF-1, and Rap1 signaling). The animal experiments also verified that HCC can improve the neurobehavioral scores and protect the neurons of IS rats, and regulate the expressions of endoplasmic reticulum stress-related targets [GRP78 (HSPA5), p-PERK (EIF2AK3), and CHOP (DDIT3)].

Data Availability
The data used to support the findings of this study are included within the article and the supplementary information files.