Eight kinds of herbs of BSHXF were provided by Yi Ling Medicine Company (Shijiazhuang, China, Drug Manufacturing Certificate: JI20150126): Polygonum Multiflorum Thunb (Product batch number: B2003001); Alpinia Oxyphylla Miq (Product batch number: B2003001); Juglans Regia L (Product batch number: B1902001); Panax Ginseng C. A. Mey (Product batch number: B1912001), Chuanxiong Rhizoma (Product batch number: B1906002), Radix Salviae (Product batch number: B1909002), Angelicae Sinensis Radix (Product batch number: B2007001) and Curcuma Aromatica Salisb (Product batch number: B1902002). All herbs in BSHXF were processed conforming to the standards of Chinese Pharmacopeia 2015. The Traditional Chinese Medicine Systems Pharmacology Database (TCMSP) contains 499 herbs and the compound ingredients of each herb registered in the 2010 edition of the Chinese Pharmacopoeia; it also includes chemicals, targets, drug-target networks, and associated drug-target-disease networks. According to the usage guide of the TCMSP database [14], the standard set for bioavailability (OB value) was $\ge$30%, and drug-like property (DL) was $\ge$0.18. We used the TCMSP database to search Panax Ginseng C. A. Mey, Angelica Sinensis, Chuanxiong Rhizoma, Curcuma Aromatica Salisb, and the practical components and targets Radix Salviae. The active chemical components of Polygonum Multiflorum Thunb, Juglans Regia L and Alpinia Oxyphylla Miq were obtained from the TCMID database. We used different databases and software sources to perform this work (Supplementary Table 1).

The TCMSP database contained the active components of five herbal medicines in BSHXF; the related targets of these active components were further searched and downloaded. We used the TCMID database to get the effective compounds of the other three herbal medicines. The associated targets of these herbals were obtained from the SwissTargetPrediction database [15]. We download all possible indicators under the condition of Homo sapiens by uploading the SMILE structural formula form for each compound, then sorted according to probability, selecting the indexes with high probability for research. Finally, we used UniProt for canonical naming.

The search terms and disease genes were screened in GeneCards [16], OMIM [17], and DisGenet databases [18]. Each drug target was compared with the disease target; common targets were screened out. The R language (Version 3.6.3, Lucent Technologies, USA) and Venny 2.1 (https://bioinfogp.cnb.csic.es/tools/venny/index.html) were used to draw the Wayne diagram; these are shown as the target genes of BSHXF in the treatment of VD.

The disease-drug gene target information was put into the STRING database to obtain the target interaction relationship. Cytoscape 3.6.0 (Version 3.6.0, National Resource for Network Biology, USA) was used to construct an optical network topology (PPI) to understand the relationship between proteins. The NetworkAnalyzer was used for network analysis. The size and degree of the nodes in the network were adjusted. The larger the graph, the greater the degree value, and the more critical the role in the network. The nodes in the network represent the targets or compounds, whereas the edges represent the associations between active components and the VD targets. According to enrichment analysis of the pathways involved in the above drug targets, the fundamental molecular mechanism of BSHXF in treating diseases was further explored, and the multi-level network correlation diagram of BSHXF-component-target was then constructed.

In the enrichment analysis of GO and KEGG, we used R language and Bioconductor. Bioconductor also contained many packages, such as clusterProfiler, org.Hs.eg.db, colorspace, stringi, DOSE, pathview, which can be used for more accurate prediction. A threshold value of p $\le$ 0.05 was used. The number of enriched genes sorted the top 20 genes in descending order, and then R language (Version 3.6.3, Lucent Technologies, USA) was used to create a bar plot and bubble graph. The main potential biological processes and signaling pathways by which BSHXF may act in the treatment of VD could be predicted through enrichment analysis. Finally, we use Cytoscape to create a visualized network.

Fifty SPF, male, Sprague Dawley rats (8 weeks, 250–280 g)were provided by the Beijing Vital River Laboratory Animal Technology Co. Ltd. (license No.: SCXK), Beijing 2016-0006. IBM SPSS was used for randomization, and the rats were grouped as follows: sham operation group (Sham group), model group (VD group), Bushenhuoxue formula group (BSHX group), autophagy activator (rapamycin) group (Rap group), and autophagy inhibitor (3-methyladenine) group (3-MA group), with n = 10/gp. The Ethics Committee of the Hebei Institute of Traditional Chinese Medicine approved the study (approval No. DWLL2019008).

The rats in each group were deprived of food for 12 h and water for 4 h before surgery. Anesthesia was 2% pentobarbital sodium (40 mg/kg, i.p.). The VD group, BSHX group, Rap group, and 3-MA group were modeled by two-vessel occlusion (2-VO) [19]. The rats were fixed in the supine position, and the skin was cut isssn the middle of the neck. The bilateral common carotid arteries and the vagus nerve were visualized, but only the common carotid artery was ligated. The wound was then sutured. In the Sham group, the vagus nerve and the common carotid artery were visualized but not ligated. Intramuscular injection of gentamicin (CSPC Ouyi Pharmaceutical Co., Ltd., China, production approval: 068180304; Specification: 2 mL = 80 mg) was administered to combat the infection.

The rats in the Rap group and 3-MA group were administered drug intracerebroventricularly 30 min before surgery. After being anesthetized with 2% pentobarbital, the rats’ heads were fixed in a stereotaxic apparatus (Stoelting, USA, 51700). A medial incision of the scalp was made to expose the skull. The coordinated used were –0.9 mm (bregma), 1.5 mm lateral, and –3.6 mm from the skull. Coordinates were marked, and small round burr holes were made with the help of a drill. Rapamycin (Shanghai Yuanye Biological Technology Co., Ltd., China. batch number: H03A10Y94335; Specification: 500 mg) was dissolved in dimethyl sulfoxide (DMSO) at a concentration of 8 ng/$\mu$L. The 3-MA (Shanghai Yuanye Biological Technology Co., Ltd., China. batch number: Y23O10C99335; Specification: 1 g) was dissolved in normal saline at a concentration of 10 mmol/L. Rapamycin solution (4 $\mu$L) or 3-MA solution (6 $\mu$L) was delivered in a microliter Hamilton syringe through the hole at a rate of almost 1 $\mu$L/min. After injection of the drug, the syringe was kept in place for 5 minutes to avoid any injected substance loss. The rat was then taken out of the apparatus, and the wound was disinfected and sutured.

On day 3 after surgery, the BSHX group received an intragastric administration of 10.14 g/kg Chinese medicine 1$\times$/day for 28 days; the remaining groups were given the same volume of normal saline every day. The composition of BSHXF was as follows: Polygonum Multiflorum, 15 g; Alpiniae Oxyphyllae, 15 g; Juglans Regia L, 15 g; Panax Ginseng C. A. Mey, 10 g; Chuanxiong, 15 g; Radix Salvia, 15 g; Angelica Sinensis, 15 g; and Curcuma Aromatica Salisb, 10 g. The solution was prepared proportionally to produce 1.014 g/mL, dissolved fully in distilled water, and refrigerated at 4 ${}^{\circ }$C.

The Morris water maze (MWM) (Beijing Zhongshidichuang Technology Co., Ltd., China, ZS-001) experiment was conducted 4 wk after surgery to test the rats’ learning and memory abilities using a spatial-orientation task. MWM is a circular tank 160-cm in diameter and 50 cm deep. The tank was divided into four imaginary quadrants. In the third quadrant, a circular platform of 12-cm diameter and a height of 29 cm was hidden 1 cm below the water surface. The water temperature was maintained at between 22 ${}^{\circ }$C and 25 ${}^{\circ }$C. From days 1 to 5, a positioning navigation experiment was conducted. The rat was placed in the water and had to find the platform from the midpoint of each quadrant. The escape incubation period refers to the latency to board the platform and stay for 5 s. The rats that did not find the platform within the specified time were recorded as 120 s. On day 6, the space exploration experiment was conducted. We removed the platform and then put the rats into the water at any of the same entry points and recorded the number of times the rats crossed the original platform within 120 seconds to test the spatial memory ability of the rats.

Three rats were taken from each group and given cardiac perfusion with 4% paraformaldehyde (Zhejiang Jinhua Top Health Biotechnology Co., Ltd. China, catalog No. 20170017). The brain was removed, fixed further in 4% paraformaldehyde, then conventionally embedded in paraffin. The wax block was trimmed to expose the hippocampus and then cut into coronal sections of 4-$\mu$m thickness. The tissue was then dewaxed and hydrated in xylene, gradient ethanol, and distilled water. After that, the tissue was stained in methylene blue for 10 min and underwent Nissl differentiation for color separation until the Nissl body was clear under the microscope. The slices were then placed into ammonium molybdate solution for 3–5 min, dehydrated, transparentized, and sealed routinely. Since the hippocampal CA1 region is most closely related to spatial discrimination and learning and memory, the pathological morphology of hippocampal CA1 in each group was observed under a digital microscope (Olympus Corporation, Japan, DP73), and the morphological changes and differences of Nissl-stained cells were observed and recorded at low and high magnification.

After anesthesia, the rats in each group were decapitated, the hippocampi were stripped off on an ice platform, the tissue blocks of about 1 mm${}^3$ were cut off with a scalpel in the CA1 region, placed in fixative for TEM (Wuhan ServiceBio Technologies Co Ltd., China, catalog No. G1102) for 2 h, washed with a phosphate-buffered solution for three times, and set in 1% osmic acid for 2 h. After cleaning three times with phosphate-buffer, the tissues were dehydrated in gradient ethanol, placed in acetone (Sinopharm Group Chemical Reagents Co., Ltd., China, catalog No. 10000418) three times, and soaked in a mixture of 1:1 and 1:2 acetone and EMBed 812 (SPI, USA, catalog No. 90529-77-4) for 1 h each. The tissue sections were then soaked in EMBed 812 for 4 h and sliced (thickness: 50 nm). After dual electron staining with uranium acetate and lead citrate, the image results were observed under TEM (HITACHI, Japan, HT7800) and recorded.

After anesthetization, the rats in each group were decapitated, and the hippocampal tissue was isolated on the ice platform and frozen in liquid nitrogen. The hippocampus was weighed, and 1 mL RIPA lysate (Shanghai Beyotime Biotechnology Co., Ltd., China, catalog No. P0013E) per 100 mg tissue was added. It was then cut into pieces, ground, centrifuged for 10 minutes, and the supernatant was taken for reserve. The protein concentration was adjusted to 6 $\mu$g/$\mu$L with lysate. The gel was prepared, and the glass plate was affixed to the electrophoresis apparatus (Beijing Liuyi Biotechnology Co., Ltd., China, DYY-III 7B). According to the quantitative results of the proteins, the total protein samples of the corresponding volume and 4$\times$ protein loading buffer was added and then denatured at 95 ${}^{\circ }$C for 10 min. The electrophoretic solution was added to the sample included in each gel hole. The voltage was adjusted to 80V for 90 min and 120V when the sample was passed through the stacking and separation gel. The target band was run to about two-thirds of the separation gel position. Electrophoresis was stopped, and the glass plate was removed. The protein bands isolated from the gel were transferred onto the PVDF membrane (Pall Corporation, USA, catalog No. T22740) by transfer electrophoresis and then incubated with a specific primary antibody (LC3 antibody: CST, USA, catalog No. 12741. P62 antibody: NSJ Bio, China, catalog No. F48010. Beclin1 antibody: Abcam, UK, catalog No. Ab210498. GAPDH: Abcam, UK, catalog No. Ab9485) diluted to a 1:800 ratio, and then horseradish peroxidase-labeled secondary antibody (Beijing Zhongshan Jinqiao, catalog No. ZB2301) was added. Quantity One software (Bio-Rad Technical Service Department, USA) was used to measure the strip’s gray value. The relative expression of the target protein was obtained by calculating the gray value ratio of the target protein and GAPDH. The relative expression of LC3 was defined as LC3-II/LC3-I ratio.

The data were analyzed and plotted using IBM SPSS 25.0 statistical software (IBM SPSS, IL, USA) and GraphPad Prism (Version 7.0, GraphPad Software, CA, USA). The results were presented as mean and standard deviation. One-way ANOVA was used to compare multiple groups that were normally distributed and had homogeneity of variance. The SNK test was used for data comparison between the two groups. A level of p $\le$ 0.05 was considered to be statistically significant for all hypothesis-testing methods.

The active ingredients of Panax Ginseng C. A. Mey, Angelica Sinensis, Chuanxiong Rhizoma, Curcuma Aromatica Salisb, and Radix Salviae were searched in TCMSP and found (22, 2, 7, 15, and 65, respectively). The active ingredients of Polygonum Multiflorum Thunb, Alpinia Oxyphylla Miq, and Juglans Regia L were collected from TCMID and found (55, 33, and 30, respectively). A total of 229 active ingredients were obtained by combining the two databases. The main active ingredients information can be found in Supplementary Table 2. Target prediction was carried out for the above ingredients. After removing the repeated components, the data obtained were 105 in Panax Ginseng C. A. Mey, 29 in Angelica Sinensis, 18 in Chuanxiong Rhizoma, 34 in Curcuma Aromatica Salisb, and 394 in Radix Salviae. The target proteins were converted to corresponding gene symbols using the Perl and UniProt databases. Target prediction with the SwissTargetPrediction was made according to the probability of $>$0.11 screening. After removing the repeated ingredients, the data points obtained were 668 in Polygonum Multiflorum Thunb, 287 in Alpinia Oxyphylla Miq and 243 in Juglans Regia L (Supplementary Table 3). The targets of the above two databases were counted, and 502 potential targets were screened.

The disease genes were screened in GeneCards, OMIM, and DisGenet databases by searching for Vascular Dementia, Ischemic Vascular Dementia, and Vascular Dementia Acute Onset. In GeneCard and DISGENet, the screening criteria were set as relevance score $>$11.38 and DSI $\ge$0.5. Finally, all the data are merged and sorted out. The information from the three databases was integrated to obtain 601 disease genes. Finally, each drug-action target was compared with the disease target, and the co-action genes of the two were screened out. The Wayne map was drawn using the R language (Version 3.6.3, Lucent Technologies, USA) and Venny 2.1 (https://bioinfogp.cnb.csic.es/tools/venny/index.html) to obtain 95 intersecting genes that might lead to VD (Supplementary Table 4).

According to the above filtering conditions, the data of the STRING database were used to draw the optical network topology diagram by Cytoscape 3.6.0 software (National Resource for Network Biology, USA) (Fig. 1A). We input the obtained files into R language for calculation and obtained the bar diagram of the core target of the protein-interaction network (Fig. 1B). The larger the number on the strip, the greater the probability of becoming the core gene. According to the BetweennessCentrality, the ClosenessCentrality, and the Degree, the top 20 core genes were selected: AKT1; IL6; GAPDH; CASP3; TNF; VEGFA; APP; MAPK1; EGFR; PTGS2; SRC; MAPK8; MYC’ NOS3; HSP90AA1; MMP9; ESR1; MMP2; ICAM1; and ACE.

Fig. 1.

The diagram of the core target of the protein interaction network. (A) The PPI visual network topology diagram of 95 common targets. The bigger the degree, the bigger the node, the darker the color, the more important the node. (B) The bar diagram of the top 30 core targets of the protein interaction network. On the left are the names of the genes. The number on the strip is the number of adjacent nodes of the corresponding gene.

The data were integrated, and Cytoscape 3.6.0 software was used to construct the visual network diagram of the “Traditional Chinese Medicine-component-target” of BSHXF-VD’s active component. According to the results of network pharmacology and the visual topology diagram, we obtained the core compounds of 8 herbs in BSHXF (Supplementary Table 5). In the diagram, different targets correspond to the same active ingredient. A target can correspond to different ingredients, thus indicating that BSHXF has multi-component and multi-target VD treatment (Fig. 2).

Fig. 2.

Network of BSHXF-component-target. The octagon represents the compound target of the drug, the triangle represents the disease, and the diamond represents the disease target.

We finally enriched 95 core targets in the GO and KEGG pathways. A total of 320 pathways were obtained, of which 158 belonged to GO processes and 162 belonged to KEGG pathways (Supplementary Tables 6,7). The first 20 pathways with p 0.05 were visualized through R language (R3.6.3). As shown in the figures, the GO biological functions of BSHXF in VD treatment involve endopeptidase activity, protein tyrosine kinase activity, growth factor receptor binding, protein serine/threonine kinase activity, G protein-coupled amine receptor activity, neurotransmitter binding. The KEGG enrichment pathway of BSHXF in the treatment of VD included the pathways of neurodegeneration - multiple diseases, PI3K-Akt signaling pathway, MAPK signaling pathway, HIF-1 signaling pathway (Figs. 3,4). Then, we used the core pathways to build the herb pathway-target network, as presented in Fig. 5.

Fig. 3.

The diagrams of GO functional-enrichment analysis. (A) The histogram of GO functional-enrichment analysis. The letters on the left of the graph are the names of GO biological functions, and the numbers below are the number of genes enriched on GO. The bar graph represents the biological functions enriched on GO, and p represents the significance of enrichment. The redder the color, the higher the degree of enrichment, and the smaller the corresponding p-value. (B) The bubble diagram of GO functional-enrichment analysis. The principle is the same as the histogram.

Fig. 4.

The diagrams of KEGG pathway enrichment analysis. (A) The histogram of KEGG pathway enrichment analysis. The letters on the left of the graph are the names of KEGG pathways, and the numbers below are the number of genes enriched on KEGG. The bar graph represents the pathways enriched on KEGG, and p represents the significance of enrichment. The redder the color, the higher the degree of enrichment, and the smaller the corresponding p-value. (B) The bubble diagram of KEGG pathway enrichment analysis. The principle is the same as the histogram.

Fig. 5.

Herb-pathway-targets network. The diamond represents the compound target of the drug, the triangle represents the pathway, and the arrowhead represents the disease target.

The results of the position-navigation experiment revealed that compared with the Sham group, the escape latency of rats in the VD group, Rap group, and 3-MA group were shown to be longer (p 0.05). Compared with the VD group, the escape latency of rats in the BSHX group was significantly shorter, and the difference was statistically significant [F (5, 49) =7.802, p = 0.000]. The results of the space-exploration experiment revealed that compared with the Sham group, the number of times the rats crossed the original platform in the VD group and Rap group was decreased, and compared with the VD group, the number of times rats crossed the original platform was increased, showing a statistically significant difference [F (5, 49) =13.62, p = 0.000] (Fig. 6).

Fig. 6.

The comparison of behavioral changes of Morris Water Maze in Rats. (A) The comparison of escape latency in rats. Compared with the Sham group,${}^{*}$p 0.05; Compared with the VD group, ${}^{\mathrm{△}}$p 0.05. (B) The comparison of the times crossing the original platform in rats. Compared with the Sham group,${}^{*}$p 0.05; Compared with the VD group,${}^{\mathrm{△}}$p 0.05.

Microscopic observation showed that the morphology and the structure of neurons in the hippocampal CA1 area of Sham group rats were as expected, the cells were orderly arranged, the nucleoli were apparent, and the number and volume of Nissl bodies were large. In the VD group and Rap group, the cells in the CA1 area were significantly damaged, with a sparse and disordered arrangement, and were irregular in shape. The nucleolus was compact or absent, and the Nissl bodies were absent. The morphology and the number of cells in the BSHX and 3-MA groups were less abnormal than those in the VD group. The cell space in the 3-MA group was enlarged and partially lost, and the Nissl bodies could be observed. The cells in the BSHX group were more orderly, and there were more Nissl bodies (Fig. 7).

Fig. 7.

Pathological morphology of hippocampal CA1 region of rats in each group. Each group had two views, one at low power (200$\times$), and the other at high power (400$\times$).

TEM observation revealed that the cells in the Sham group appeared normal in morphology and complete in structure, but no autophagy was observed. The VD group and Rap group cells were characterized by abnormal cell structure, nucleolus shrinkage and deformation, reduced mitochondrial number, and endoplasmic reticulum fracture. Autophagic lysosomes were observed intracellularly. The BSHX group and 3-MA group’s cell morphology was better than that of the VD group. The organelle was relatively complete, the chromatin distribution was more uniform, and the autophagosomes were rare. The overall degree of cell damage from heavy to light was as follows: the VD group, the Rap group, the 3-MA group, the BSHX group, and the Sham group (Fig. 8).

Fig. 8.

Neuronal damage and autophagy in the hippocampal CA1 region of rats in each group were observed by TEM. Each group had two views, one at low power (2500$\times$), and the other at high power (6000$\times$). In the Sham group, the nuclei (N) were large and round, with abundant organelles, slightly damaged mitochondria(M), abundant secondary lysosomes (SLy). Rough endoplasmic reticulum (RER) was slightly dilated, surface ribosomes were degranulated, Golgi bodies (GO) were abundant, and glycogen (Gl) was evenly dispersed in the cells. No autophagy was observed in the cells (0). In the VD group, the nuclei were abnormal, the cells were irregular, there was more lipofuscin (Lip) and secondary lysosome accumulation in the cells, mitochondria were slightly vacuolated, and autophagic lysosomes (ASS) were observed intracellular (2). In the BSHX group, the nuclei were large and translucent, with abundant organelles, moderate mitochondrial edema, and no autophagy was observed (0). The RAP group had abnormal nuclei, irregular cells, more lipofuscin in cells, slight mitochondrial vacuolation, and more autophagy (3). In the 3-MA group, the nuclei were abnormal, and the cell body was slightly irregular. There was a certain amount of lipofuscin and a small number of secondary lysosomes in the cells. Mitochondria were slightly damaged, and no autophagy was observed (0).

The experiment was repeated 3 times, and there were no differences. Compared with the Sham group, the expression levels of Beclin-1 and LC3 in the VD group and Rap group were increased, whereas the expression of P62 was decreased, and the difference was statistically significant [F (5, 19) = 18.064, p = 0.000]. Compared with the VD group, the expression of Beclin-1 and LC3 in the BSHX group and 3-MA group was lower. In contrast, the expression of P62 was greater, showing a statistically significant difference [F (5,19) =12.699, p = 0.000] (Fig. 9).

Fig. 9.

Comparison of the expression levels of autophagy-related protein in rats. (A) The diagram of the western blot. (B) The diagram of expressions of autophagy-related protein. [Mean (SD), n = 4]. Compared with the Sham group,${}^{*}$p 0.05; Compared with the VD group,${}^{\mathrm{△}}$p 0.05. The red columns represent Beclin1/GAPDH, and the green columns represent LC3-II/LC3-I and P62/GAPDH.