In this study, UHPLC-LTQ-Orbitrap-MS/MS method was used to identify the components of DHI. The reagents and instruments used were as follows: DHI, provided by Shandong Danhong Pharmaceutical Co., LTD., specification 10 ml/piece. Ultimate 3000 ultra performance liquid chromatography and LTQ Orbitrap XL mass spectrometer (equipped with an electrospray ion source (ESI) and xcalibur 2.1 Chemstation) were purchased from Thermo Scientific, United States. Mass spectrometric conditions: positive and negative ion detection mode, electrospray ion source (ESI), sheath gas and auxiliary gas were nitrogen gas with purity of >99%; Sheath gas flow rate 40 U, auxiliary gas flow rate 20 U; Collision gas was helium with purity of >99.99%; Ionization source voltage, 3.0 kV; Ion source temperature: 350°C; Drying gas flow rate 15 L·min-1; Collision voltage: 6–10 V; First order mass spectra were obtained by Fourier transform with high-resolution and full scan (FT) with a mass ranging from M/Z: 50–1200; Detection resolution: 3,000; Liquid phase conditions were: flow rate: 0.200 ml/min; Needle wash: methanol; Column oven temperature: 30°C; Autosampler injection volume: 4.00 μl; Mobile phase conditions were 0.1% formic acid (a), acetonitrile (b): 0.01–5.00 min B: 15–20, 5.00–20.00 min B: 20–35, 20–40 min B: 35–55, 40–45 min B: 55–35, 45–55 min B: 55–15. The column was: waters HSS T3 UPLC C18 column (1.7 µm, 2.1 × 100 mm).

In addition, PubMed (https://pubmed.ncbi.nlm.nih.gov/) and CNKI (https://www.cnki.net/) databases were used to screen the main compounds of DHI. Then, PubChem (https://pubchem.ncbi.nlm.nih.gov/) database was used to find the SMILES structural information of candidate compounds (Kim et al., 2016). At the same time, ChemDraw (http://www.chemdraw.com.cn/) software was used to draw the 2D structure of the candidate compounds (Evans, 2014).

In this study, the following three databases were mainly used to find known or predicted targets related to compounds: SuperPred (http://prediction.charite.de/) (Nickel et al., 2014), SwissTargetPrediction (http://www.swisstargetprediction.ch/) (Gfeller et al., 2014) and BATMAN-TCM (http://bionet.ncpsb.org/batman-tcm/) (Liu et al., 2016). In addition, the UniProt (http://www.uniprot.org/) online database was used to convert the obtained protein names into gene names, and the source species was restricted to “Homo sapiens,” so that the names were standardized for later processing of the data (Liu et al., 2019).

The human gene data associated with AMI are from the following six databases: DigSee database (http://210.107.182.61/geneSearch/) (Kim et al., 2017), DisGeNet database (http://www.disgenet.org/search) (Piñero et al., 2017), OMIM database (https://omim.org/) (Amberger et al., 2015), TTD database (https://db.idrblab.org/ttd/) (Chen, 2002), MalaCards database (https://www.malacards.org/) (Rappaport et al., 2017), and GEO database (https://www.ncbi.nlm.nih.gov/gds/) (Barrett et al., 2012). The duplicate values of human gene targets related to AMI obtained from the above six databases were deleted for further analysis.

In this study, the AMI targets and DHI compound targets were crossed and the same targets were uploaded to the STRING 11.0 database (Szklarczyk et al., 2017). In addition, the species source was “Homo sapiens,” and the confidence score was set to be greater than 0.7.

In this study, Cytoscape 3.7.1 (https://cytoscape.org/) was used to construct the following six visualization networks: 1) compound-target network; 2) compound-AMI target network; 3) PPI network of the common target of DHI compound and AMI, namely DHI-AMI target protein-protein interaction network; 4) modular analysis network; 5) DHI-key compound-key target-pathway network. In addition, the NetworkAnalyzer tool in Cytoscape was used to perform a topological analysis of the network, including “degree,” “betweenness” and “closeness.” The higher the value of these parameters, the more important the node is in the network. Moreover, on this study, the MCODE plug-in in Cytoscape software was used to analyze the PPI network (Chen et al., 2021). The default parameters are: Degree Cutoff = 2; Node Score Cutoff = 0.2; K-Core = 2; Max. Depth = 100.

To better understand the biological functions and signal transduction pathways of the potential core targets of DHI in the treatment of AMI, the “Bioconductor package” of R 3.6.1 software and Kyoto encyclopedia of genes and genomes (KEGG) was used for the for Gene Ontology (GO) and pathway enrichment analysis, respectively (Robinson et al., 2009).

The PubChem database platform was used to download the main compound files in SDF format, and the ChemDraw software converted the SDF format to mol2 format. Afterward, Autodock Tools 1.5.6 software was used to load the processed mol2 format files, set up rotatable bonds, etc., and finally, save them as PDBQT format files (Morris et al., 2009). This study screened the protein conformations of key targets from RCSB PDB (https://www.rcsb.org/) (Goodsell et al., 2020). The screening conditions are: 1) the protein structure obtained by the X-ray crystal diffraction method; 2) the crystal resolution of the protein is less than 3Å; 3) the biological source of the protein should be human; 4) preferentially select the protein structure reported in the literature on molecular docking. At the same time, Notepad++ (https://notepad-plus-plus.org/) and AutoDockTools were used to remove water molecules and co-crystallized original ligand molecules, adding hydrogen and charge, merge non-polar hydrogen, and determine the location of active pockets. Then, Autodock Vina 1.1.2 was used to perform molecular docking calculations with the processed ligands and receptors (Trott and Olson, 2009). Finally, PyMol 2.3.2 (https://pymol.org/2/) was used to visualize the molecular docking results (Yuan et al., 2016).

DHI, provided by Shandong Danhong Pharmaceutical Co., LTD., specification 10ml/piece. Nitrotetrazolium chloride blue (N-BT), Amresco Co, Lot 2541C012. Tp-213 electronic balance, Beijing Sartorius Instrument System Co., LTD. Dt-2000 electronic balance, American Shuangjie Brothers Co., LTD. Mpids-500 Color Pathological Image Analysis System, Beijing Air Sea Company; Canon EOS 5D Mark ⅲ, Canon China Co., LTD. Vevo 3100 High resolution Ultrasound Imaging System for Small Animals, Visual Sonics Inc.

A total of 80 male SD rats weighing 180–200 g were purchased from Scxk (Beijing) Biotechnology Co., Ltd. (animal qualification certificate no.: SCXK (Beijing) 2019-0010). Rats were maintained on a 12 h light/dark cycle, at 22 ± 2°C and controlled humidity. Besides, they had ad libitum access to water and rodent food. Animal care, surgery, and handling procedures were performed in accordance with the regulations of the Ministry of Science and Technology of the People’s Republic of China ([2006] 398) and were approved by the Animal Care Committee of Xiyuan Hospital of the China Academy of Chinese Medical Sciences (SYXK [JING] 2019-0010).

Rats were anesthetized by intraperitoneal injection of 12 ml/kg 2% sodium pentobarbital and fixed in the supine position on a plate in. Then, a thoracotomy was performed at the fourth and fifth intercostal points on the left side of the rat, and the pericardium was opened. The thoracic cavity was gently pressed to extrude the heart, and the left anterior descending coronary artery was ligated with a 0 suture approximately 2 mm below the left atrial appendage. The heart was immediately returned to the thoracic cavity, and the opened skin was sutured. At the same time, rats in the sham group only thread and did not ligate.

After successful modeling, rats were randomly assigned to different groups such as sham (n = 20), model (n = 30), and DHI 6 ml/kg (n = 30). Rats in each group received a continuous intraperitoneal injection for seven consecutive days, and the sham group and the model group received normal saline in the same manner.

A small animal ultrasound instrument was used for cardiac ultrasonography, and echocardiography was performed 3 h after the last administration. After anesthesia, rats were fixed in the supine position on a thermostatic heating plate, and their limbs were connected to electrocardiogram electrodes for monitoring heart rate and recording the electrocardiogram. Cardiac ultrasound was performed using a Vevo 3100 small animal ultrasound instrument. After the chest hair was removed, the body of the rats was tilted 30° to the left and an ultrasound coupling agent was applied. The ultrasonic probe is placed on the left side of the sternum at an angle of 10°–30° to the midline of the sternum, showing the long axis section of the left ventricle of the sternum. Apply M-mode ultrasound to the long axis view of the left ventricle, run the sampling line through the left anterior and posterior sidewall at the level of the chordal chordae of the mitral valve, record and measure the motion curve of the left ventricle. Then the probe is rotated 90° clockwise, making moderate adjustments to show that the left ventricle is short. Using anatomical M-mode ultrasound on the short-axis section of the left ventricular papillary muscle, the sampling line passes through the anterior septum and posterior wall. Guided by two-dimensional images, the m-shaped curve of left ventricular motion was recorded and measured, including left ventricular ejection fraction (EF), left ventricular short axis shortening rate (FS%), and left ventricular anterior wall thickness (LVAWd and LVAWs) in the end diastolic and end systolic stages.

Three hours after the last administration to the rats of each group, the heart was immediately removed, the residual blood in the cardiac cavity was flushed with normal saline, the excess fluid was absorbed with filter paper, and then weighed. The ventricle was cut evenly from the sulcus parallel to the coronary artery under the ligation line into five equal thickness slices. Then, these slices were weighed separately, placed in 0.2% nitrotetrazolium blue dye solution (N-BT), and stained for 2–3 min in the dark at room temperature. Subsequently, the camera was used to photograph the anterior and posterior sides of five myocardiums to obtain images, and the pathological color image analysis system was used to measure each piece on both sides of the myocardial infarct size (N-BT non-stained area) and non-infarct size (N-BT stained area). At the same time, total ventricular muscle size, total infarct size, and percentage of infarct size to ventricular size and whole heart size were also calculated.

Five rats from each group were selected for cardiac histopathological observation. The method was as follows: after anesthesia, the hearts of the rats were removed, the right ventricle was stripped and discarded, and the myocardial tissue in the infarct junction area was collected. Cardiac tissue specimens were placed in a prepared 4% formaldehyde fixative. After dehydration, they were conventionally embedded in paraffin and sectioned. Then the tissues were stained with HE and Masson. The degree of changes in the myocardium was observed under the microscope.

RIPA lysate (Thermo Scientific) was used to extract protein samples, and protein concentration (Pierce) was determined by the BCA method. The SDS-PAGE gel was prepared using the Mini-PROTEAN Tetra hand filling system (Bio-rad), and each well was loaded with 30 μg normalized final loading concentration. After electrophoresis, the protein was transferred to the PVDF membrane (Millipore), and the corresponding antibody was incubated to achieve immunoaffinity between antibody and antigen. Developing with ECL (Thermo Scientific), and exposing with Kodak Carestream to obtain protein imprinted bands.

Data are expressed as “mean ± standard deviation ( x ¯ ±s),” and SPSS 25.0 statistical analysis software is used for data analysis and processing. For data conforming to the normal distribution, one-way analysis of variance (ANOVA) is used for two or more groups. T Homogeneity of variance was tested using Student-Newman-Keuls and the uneven variance was tested using Tamhane’s T2; data that did not conform to the normal distribution was tested using a non-parametric test. Differences with p < 0.05 were considered statistically significant.

After the compound screening, a total of 12 major compounds were incorporated into this study (Zhao et al., 2017; Li et al., 2019; Xu et al., 2019), including Salvianolic acid A, Salvianolic acid B, Danshensu, Protocatechuic acid, Rosmarinic acid, and Caffeic acid, which were also detected in UHPLC-LTQ-Orbitrap-MS/MS analysis (Figure 2 and Supplementary Tables S1, S2).

By searching the SuperPred, SwissTargetPrediction, and BATMAN-TCM databases, a total of 372 targets were found to be associated with the compound. The compound-target network of DHI is shown in Figure 3A. It consists of 384 nodes (12 compound nodes, 372 target nodes) and 708 edges. This network shows that a single target can be regulated by multiple compounds, which could play a vital role in the treatment of AMI. Indeed, a single compound may act on multiple potential targets.

In this study, mainly AMI-related targets were retrieved from the five databases DigSee, DisGeNet, OMIM, TTD, and MalaCards, and the AMI differential genes of published articles were combined and then duplicate data were detected (Guo et al., 2021b). A total of 1929 potential targets related to AMI were obtained. At the same time, the obtained AMI potential targets and compound targets were intersected and uploaded to Cytoscape to construct a compound-AMI target network (Figure 3B). As shown in Figure 3C, this network contains 170 nodes (12 compound nodes, 158 target nodes) and 318 edges. In addition, the topological characteristics of the network were analyzed. The results show that the top three compounds in terms of “degree,” “betweenness” and “closeness” are ferulic acid, caffeic acid and salvianolic acid A, respectively. And the three largest targets with topological parameters are carbonic anhydrase 9 (CA9), carbonic anhydrase 2 (CA2) and aldo-keto reductase family 1 member B1 (AKR1B1). These compounds and targets may play an important role in the treatment of AMI with DHI.

In this study, the AMI target and the DHI target were crossed and imported into the STRING database platform to construct a DHI-AMI target protein-protein interaction network to better understand the complex interactions between common targets. As shown in Figure 4A, the network contains 151 nodes and 859 edges; the node’s size is proportional to the degree value. In this study, the top ten nodes with a median network value were selected as important nodes: APP, MAPK1, TNF, MAPK8, STAT3, EGFR, MAPK3, PIK3CA, JUN, SRC (Supplementary Table S3).

Traditional Chinese medicine has the complex characteristics of multi-component and multi-target, and the module analysis based on the “law of similar attraction” is considered one of the most powerful methods to explain the mechanism of Chinese medicine (Nie et al., 2020). A network module means that the connection between the nodes of the sub-network is closer than the connection with the rest of the network, and it has the characteristics of high interconnection, which helps to find the structure and function information of similar proteins in the network (Guimerà and Nunes Amaral, 2005). In this study, a modular analysis of the DHI-AMI target protein-protein interaction network was performed. The first three modules with MCODE scores greater than 9.00 were finally selected as important modules (Figures 4B–D). Ten potential key targets are mainly enriched in the first two modules; namely, PIK3CA and APP in module 1; MAPK1, TNF, MAPK8, STAT3, EGFR, MAPK3, and JUN in module 2. SRC is not enriched in important modules.

In this study, GO enrichment and KEGG pathway enrichment analysis were performed for the DHI-AMI target protein-protein interaction network. The three important modules, in order to better understand the cellular component (CC), molecular function (MF), biological process (BP), and signaling pathways associated with the treatment of AMI with DHI. According to the corrected p-value (p < 0.05), this study only displays the top 10 GO elements in the legend. The results of GO functional enrichment analysis showed that protein-protein interaction targets were mainly related to regulation of blood vessel size, blood circulation, cell membrane, and enzyme activity (Figure 5A); module 1 was mainly related to G protein-coupled receptors, blood vessel diameter, vasoconstriction, intracellular calcium ions, synapses, G protein-coupled amine receptors, and phospholipases (Figure 5C); module 2 was mainly related to oxidative stress response, cell membrane, phosphatase binding, and MAP kinase activity (Figure 5E); module 3 was mainly related to the regulation of blood circulation, dendritic cell chemotaxis, signaling receptor activity, outer plasma membrane, C-C chemokines, and G protein coupling (Figure 5G).

Further, KEGG pathway enrichment analysis was performed for 151 protein-protein interaction targets, and a total of 153 pathways were screened (adjusted p-value < 0.05). In this study, only the top 30 pathways are shown in the legends. The results of the KEGG pathway enrichment analysis show (Figure 5B) that the protein interaction targets are mainly enriched in the calcium signaling pathway, IL-17 signaling pathway, TNF signaling pathway, and other signaling pathways. These pathways are mainly related to endocrine system, signal transduction and immune system.

Through the KEGG pathway enrichment analysis of 16 targets in module 1, a total of seven signaling pathways were screened (adjusted p-value < 0.05), which are mainly enriched in neuroactive ligand-receptor interaction, calcium signaling pathway, and other signaling pathways. These signaling pathways are mainly related to signal transduction, nervous system and immune system (Figure 5D). Through the KEGG pathway enrichment analysis of 21 targets in module 2 (adjusted p-value < 0.05), a total of 141 signaling pathways were screened, which are mainly enriched with toxoplasmosis, AGE-RAGE signaling pathways in diabetic complications, IL-17 signaling pathways and so on. These signaling pathways are mainly related to infectious diseases, cancer, immune system and endocrine system (Figure 5F). Through the KEGG pathway enrichment analysis of 11 targets in module 3 (adjusted p-value < 0.05), a total of eight signaling pathways were screened, which are mainly enriched in the interaction of viral proteins with cytokines and cytokine receptors, interaction between cytokine and cytokine receptors and other signaling pathways. These pathways are mainly related to signaling molecules, signal transduction and the immune system (Figure 5H).

Molecular docking was used to explore the binding methods of ten key targets and corresponding compounds. The crystal structures of 10 key targets were retrieved and downloaded from PDB database and saved as PDB format file. In addition, the 3D structures of the corresponding compounds (caffeic acid, ferulic acid, rosmarinic acid, salvianolic acid B, danshensu, cytidine, salvianolic acid A) were downloaded from PubChem database. The molecular docking of four compounds with EGFR was analyzed as an example. As shown in Figure 6, caffeic acid forms three hydrogen bonds with residues PHE-856, LYS-745, and CYS-775 on EGFR protein; danshensu forms three hydrogen bonds with residues LEU-777, PHE-856 and LYS-745 on EGFR protein; ferulic acid forms three hydrogen bonds with residues ALA-743, PHE-856 and LYS-745 on EGFR protein; rosmarinic acid forms four hydrogen bonds with MET-793, CYS-775, PHE-856 and ASP-855 residues on the EGFR protein. Interestingly, all four compounds shared some similar binding sites. In addition, the docking scores of all ligand and receptor combinations in this study were greater than −5.520 (Hsin et al., 2013), indicating that the key target protein and corresponding compounds have good binding affinity (Table 1).

In this study, Cytoscape software was used to construct a DHI-key compound-key target-pathway network to systematically reveal the mechanism of DHI in the treatment of AMI. As shown in Figure 7A, the DHI-key compound-key target-pathway network includes 47 nodes (one drug node, seven compound nodes, 10 core target nodes, and 29 pathway nodes) and 194 edges. In addition, the calcium signaling pathway was also significantly enriched (Figure 7B).

Compared with the sham group, end-systolic and end-diastolic anterior wall thicknesses (LVAWs and LVAWd) of the model group were significantly decreased, whereas ejection fraction (EF) and left ventricular fractional shortening (FS) were significantly declined (p < 0.01). In addition, the LVAWs, EF, and FS of the DHI group were significantly increased compared with the sham group (p < 0.05 or p < 0.01) (Figures 8A–C and Tables 2, 3).

The results showed that the percentage of myocardial infarction area to ventricular area and total heart area was significantly lower in the sham group than in the model group (23.21 ± 6.14% and 14.97 ± 4.46%, p < 0.01). In addition, the percentage of myocardial infarction size to ventricular size and total heart size was significantly lower in the DHI 6 mg/kg group than in the model group (11.21 ± 4.06% and 7.43 ± 2.56%, p < 0.01) (Figures 8D,E and Table 4).

The results showed that the myocardium in the sham group was intact, the muscle fibers were neatly arranged, and the shape was as usual. There was no degeneration or necrosis. Also, there was no congestion or hemorrhage in the interstitium, no obvious infiltration of inflammatory cells, and no deposition of collagen fibers. However, the pathological findings of the model group showed disordered myocardial fiber arrangement, widened muscle fiber spacing, focal edema and degeneration of myocardial cells, necrosis of individual cells, interstitial vascular proliferation, hyperemia, edema, large number of inflammatory cell infiltration, and extensive deposition of collagen fibers. It is worth noting that the pathological morphology of the myocardial tissue in the DHI 6 mg/kg group showed an orderly arrangement of myocardial fibers, no cell necrosis in the interfibrous septum, a small number of myocardial cells with mild edema and degeneration, a small number of inflammatory cells scattered in the interstitium, and a small area of collagen fiber deposition (Figures 9A,B).

The results showed that there was no significant difference between each group in the expression of PLB. Compared with the model group, the expression of CaMK II was significantly downregulated (p < 0.01), and the expression of SERCA was significantly upregulated in the DHI group (p < 0.01). Compared with the sham group, CaMK II expression was significantly upregulated (p < 0.01), and SERCA expression was significantly downregulated in the model group (p < 0.01) (Figures 9C–E and Table 5).