DengZhanXiXin injection (specifications: 5.32 mg scutellarin, 2.26 mg 3,4-O-dicaffeoylquinic acid, 1.10 mg 3,5-O-dicaffeoylquinic acid, 1.79 mg erigoster B, 2.70 mg 4,5-O-dicaffeoylquinic acid, and 11.26 mg erigeroster, 10 ml/ampoule, Lot No. 20180137) was provided by Yunnan Biovalley Pharmaceutical Co., Ltd (Kunming, China).

The animal experiments were conducted under the direction of NIH Guidelines for the Care and Use of Laboratory Animals (NIH Publications No. 80–23, revised 1996). The procedures were approved by the Animal Care and Use Committees of Beijing Normal University, China. Adult male Sprague-Dawley rats (270 ± 10 g) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. The animals were allowed to acclimate for 7 days before the experiments. Rats were randomly divided into a sham-operated (sham) group, MCAO model group, and DZXI-treated (DZXI) group (n = 10 rats each group). Animals were housed in cages in a controlled environment (22–25°C, 50% humidity, and a 12-h light/dark cycle) with free access to standard laboratory chow and distilled water. All efforts were made to minimize animal suffering and the number of animals used.

All rats were fasted overnight before surgery but allowed free access to water. The rats were anesthetized with 3.5% chloral hydrate (350 mg/kg body weight, i.p.), and MCAO surgery was performed. Focal cerebral ischemia was induced using the filament model as described in previous research (Mhairi Macrae, 1992). Briefly, the right common carotid artery (CCA) and external and internal carotid arteries (ECA and ICA) were exposed through a midline cut. Temporarily clamp the CCA with an arterial clip, ligate the proximal bifurcation of the ECA and the proximal end of the CCA, then clamp the ICA with an arterial clip, and make a small incision between the CCA ligation point and the clamping point. A nylon monofilament coated at the tip with 5 mm of silicone (diameter 0.36 ± 0.02 mm) was inserted into the lumen of the right ICA through the CCA. Keep inserting until encountering slight resistance, then the tip reaches the entrance to the middle cerebral artery (MCA). The length of the inserted filament was around 18–22 mm from the carotid bifurcation point. Then, the CCA was ligated to fix the filament and prevent bleeding. After a 90-min ischemic period, the occluding filament was gently withdrawn to restore the blood flow for reperfusion injury, followed by surgical site suturing. The rectal temperature was maintained at 37 ± 0.5°C with a thermostatically controlled heating blanket throughout the surgical procedure. In the sham group, animals were subjected to the same procedure, except for the insertion of the nylon monofilament.

DZXI (0.18 mL/100 g) was intravenously injected into rats in the DZXI group. The concentrations of main active compounds (in 10 ml/ampoule): 5.32 mg scutellarin, 2.26 mg 3,4-O-dicaffeoylquinic acid, 1.10 mg 3,5-Odicaffeoylquinic acid, 1.79 mg erigoster B, 2.70 mg 4,5-O-dicaffeoylquinic acid, and 11.26 mg erigeroster. The first drug treatment was conducted immediately after the MCAO surgery, and subsequent treatments were continually performed at an interval of 12 h, i.e., twice a day. The drug treatments lasted for 7 days. The dosage and treatment course referred to general usage for humans and the equivalent dose ratio based on body surface area (An et al., 2021). The rats in the sham group and model group were treated with the same volume of saline in the same way.

Modified neurological severity score (mNSS) test was conducted by raters who were blind to animal grouping to measure neurological function at 6 h, 24 h, and 7 days after MCAO and DZXI treatment. The mNSS test is the standard and globally accepted method to evaluate the severity of post-stroke injury and recovery. The mNSS test comprised motor, sensory, reflex, and balance tests. The total score of mNSS ranges from 0 to 18, and a higher test score indicates a more severe neurological deficit (Chen et al., 2001).

Infarction caused by MCAO was confirmed by TTC staining to assess infarction size (Bederson et al., 1986). The rats were sacrificed 7 days after MCAO and drug treatment. The rat brains were rapidly removed and frozen at −20°C for 30 mi, after which 2 mm thick coronal sections of the rat brains were cut. The brain sections were incubated with 1% TTC in the dark for 20 min at 37°C. After TTC staining, sections were fixed in 4% polyformaldehyde buffered solution for 20 min. The cerebral infarct area was outlined in white in MCAO rats.

Pathological alterations in the morphology of neural cells around the focal ischemic area, also referred to as the ischemic penumbra, were evaluated by hematoxylin-eosin (HE) staining. After the rats were anesthetized with 3.5% chloral hydrate, they were perfused with warm saline via the left ventricle, and their brains were fixed with 2% glutaraldehyde and 4% paraformaldehyde (PFA). Then the whole brain was embedded in paraffin and serially coronally sectioned. The sections were dewaxed with xylene, dehydrated with a gradient of alcohol solutions, and washed with running water. After this, the brain sections were stained with haematoxylin and differentiated with 0.5% hydrochloric acid alcoholic solution, then washed with running water. Then, the sections were returned to a blue color by incubating them with a saturated lithium carbonate solution for 1 min and stained with a 0.1–0.5% eosin solution for 10 min. The sections were examined using a light microscope and then photographed. Moreover, the numbers of dead neural cells in sections of rats in each group were counted using ImageJ software. The ratio was calculated as the number of dead neural cells divided by the number of total neural cells in images. Notably, histological staining was accomplished by raters who were blind to the animal grouping situation.

In the sham, model, and DZXI groups, rats were sacrificed 7 days after surgery, and ischemic hemispheres (four replicates in each group) containing the ischemic core were harvested. The samples were digested with trypsin and labeled with TMT (tandem mass tags, Thermo). An equal amount of each labeled sample was mixed, chromatographically separated, and finally subjected to an LC-MS/MS analysis. The peptide mass maps were analyzed using Proteome Discoverer^TM 2.2 software (Thermo) and the UniProt database. The significantly differential proteins between the two groups were identified with a fold change (FC) >1.2 and p-value (calculated by t-test) < 0.05. These differentially expressed proteins were entered as foci into the subsequent biological information analysis.

To interpret experimental information and discover potentially valuable proteins, the proteomic data were analyzed using biological information from the Gene Ontology (GO) and STRING databases. GO enrichment analysis of 191 differentially expressed proteins was performed on the Metascape platform to classify these proteins according to their biological process (Zhou et al., 2019). GO enrichment analysis refers to the distribution of experimental data, which was compared with the distribution of the overall protein, confirming experimentally identified proteins were significantly enriched in which categories. Subsequently, to further understand the biological context of the differentially expressed proteins, protein interaction analysis was carried out using the free website search tool STRING11.5. The STRING database (https://string-db.org) systematically collects and integrates known and predicted protein-protein interaction data, both physical interactions and functional associations, for many organisms (Szklarczyk et al., 2023). The gene symbol list of 191 differentially expressed proteins was input into the STRING database to identify known and predicted protein-protein interaction networks.

The potential binding capacity between the main compounds of DZXI and key protein targets was analyzed by molecular docking using the Autodock Vina and AutoDock (Trott and Olson, 2010). The 3D structures of the main compounds were obtained from the TCMSP database, whereas the key protein targets' 3D structures were obtained from the RCSB Protein Data Bank (PDB) and AlphaFold. The figures of the active binding sites between chemical compounds and key proteins were generated with the PyMOL software (Burley et al., 2017).

Neurobehavioral data were analyzed using two-way ANOVA, followed by Turkey's HSD test to complete post-hoc multiple comparisons. Statistical figure was drawn in Prism 8.0 with data expressed as mean ± SD. The difference in infarction ratio and dead cell ratio between groups was analyzed using one-way ANOVA and Turkey's HSD test, then plotted in R. Differential analysis for proteomic data was calculated by Student's t-test with FDR correction in R. The visualization of protein expression data was conducted in R. The criterion for statistical significance was set as p < 0.05.

To evaluate the potential neuroprotection activity of DZXI in focal cerebral ischemia, rats were subjected to MCAO surgery. The severity of neurological deficits, such as hemiparesis and motor coordination, is a crucial issue for the evaluation of stroke consequences. After 6 h, 24 h, and 7 days of treatment, the post-stroke neurological deficits of the experimental rats in each group were scored using the mNSS scoring criterion. The results of mNSS showed that DZXI intervention could significantly improve the neurological function recovery of ischemia-reperfusion injury rats compared with MCAO rats (Figure 1A). Besides, MCAO was validated by TTC staining, where infarction in the right caudate putamen and temporal lobe cortex could be observed 7 days post-MCAO. Compared with the typical image of the sham group, the infarct size of the MCAO group was significantly increased after MCAO injury. When treated with DZXI, TTC staining images showed that the infarct volume in the DZXI group significantly reduced compared with the MCAO group (Figures 1B, C).

Besides, we examined the morphology of cortical neural cells in ischemic penumbra through HE staining. In the brain tissue of sham group, the morphology of neural cells was normal, and no pathological degeneration or necrosis of nerve cells was observed (Figure 2A left). In contrast, in the MCAO group, the penumbra tissue cells were arranged irregularly and appeared to have more cellular swelling and necrosis with a certain degree of fuzzy neural structure and deep nuclear staining (Figure 2A middle). Moreover, compared with the sham group, the number of dead neural cells stained by HE staining increased in the cortex of MCAO rats (Figure 2B). After DZXI treatment, the cortical penumbra neural cells exhibited restored intact morphological features, including more orderly arrangement and clearer cell structures (Figure 2A, right). Furthermore, the number of dead neural cells in the cortical penumbra after DZXI treatment significantly declined compared with the MCAO group (Figure 2B). These results illustrated that the DZXI treatment could protect neural cells from ischemic injury, possibly associated with its ameliorative effect on IS.

Through the proteomic detection and analysis, we identified 145 differentially expressed proteins (DEPs) between the sham and MCAO model groups and 191 DEPs between the model and DZXI groups after 7-day treatment (Supplementary Table S1). The result of clustering analysis for these proteins is depicted in Figures 3A, B. The GO analysis of DEPs between the sham group and MCAO model groups showed that ensheathment of neurons, response to oxidative stress, response to carbon dioxide and response to wounding were significantly enriched biological processes items (Figure 3C), whereas response to metal ion, neurofilament bundle assembly and modulation of chemical synaptic transmission were most enriched items of DEPs between the model and DZXI groups (Figure 3D). The GO enrichment analysis illustrated that the protein changes caused by acute ischemic injury primarily reflected the large number of cell deaths that leads to significant reduction in protein components involved in neuronal myelin formation, as well as the response to ischemia and hypoxia injury (such as oxidative stress, response to carbon dioxide, and response to injury). Moreover, the protein changes due to DZXI treatment mainly concentrated on proteins that respond to metal ions (such as calcium dependent proteins), proteins involved in neurofilament bundle assembly, and proteins related to trans-synaptic signaling and chemical synaptic transmission (such as chemical neurotransmitter receptors), suggesting that these protein clusters could be potentially influenced by DZXI treatment.

Furthermore, we selected and categorized 39 principal proteins among DEPs between the MCAO model and DZXI groups after 7-day treatment based on their specific functions. The detailed information of these proteins, including accession number, protein name, gene symbol, MW, pI, protein score, and fold change, is summarized in Table 1, and their specific functions are displayed in Table 2. The protein expression levels of 23 proteins were decreased, and 16 proteins were increased after DZXI treatment. These proteins were divided into 13 categories according to their specific functions, such as anti-inflammatory, calcium-dependent phospholipid binding, chemical synaptic transmission, cell signaling, energy metabolism, and antioxidation. Among these proteins, eight cell signaling-associated proteins and six chemical synaptic transmission-associated proteins were impacted; the other 25 proteins are involved in anti-inflammatory, calcium-dependent phospholipid binding, energy metabolism, and antioxidation. Therefore, cell signaling and chemical synaptic transmission were the most affected physiological processes during DZXI treatment. As shown in Supplementary Table S2, the molecular function and biological process categories of these 39 proteins were determined based on biological function.

STRING analysis was performed to construct a protein-protein interaction (PPI) network. One hundred eighty-eight DEPs between the MCAO model and DZXI groups were effectively recognized for PPI network analysis based on the STRING database to build a high-quality PPI network. The comprehensive PPI regulation network is shown in Figure 4. This network contains 188 proteins and 276 connected edges, with an average node degree of 2.94, and the average local clustering coefficient is 0.388, indicating that the relationship between nodes in this network is relatively close. We further extracted the specified function cluster network focused on cell signaling and chemical synaptic transmission (Figure 5). The functional sub-network is generated from the 15 DEPs, and nine of them are involved in cell signaling, while six of them are relevant to chemical synaptic transmission. This sub-network comprises 15 nodes and 31 edges, its average node degree is 4.13, and the average local clustering coefficient is 0.642, representing the verified complicated and tight interactions between these proteins. The network analysis proves that DZXI may regulate cell signaling and synaptic transmission during IS pathological processes.

To calculate potential binding capacity between five DZXI active compounds [including scutellarin, 3,4-O-dicaffeoylquinic acid, 3,5-O-dicaffeoylquinic acid, 4,5-Odicaffeoylquinic acid, erigoster B which are highly exposed systematically in vivo (Guo et al., 2018; Huang et al., 2019)] and key target proteins, we performed computational simulations for each combination, respectively. The structures for each compound and protein were collected from the corresponding database. AutoDock was applied to conduct molecular virtual docking and calculate binding free energy. The binding energy was used as the reference to predict drug-target binding affinity. It is generally believed that a binding energy <0 indicates that tested molecules prefer to form a stable complex from spontaneous binding. Moreover, the lower binding energy represents a more stable complex formed and a greater possibility of combination occurring. Generally, the binding energy below −5 kcal/mol is considered a moderately tight docking affinity (Galma et al., 2021). Furthermore, a prior work using virtual screening to find a natural inhibitor for cAMP and cAMP-inhibited cGMP3′5′-cyclic phosphodiesterase 10A (PDE10A) accepted binding energy lower than−6 kcal/mol as sufficient affinity for the target (Al-Nema et al., 2018). Therefore, the present simulation indicated that each key target protein and DZXI active compound can form a relatively stable complex, because all their binding energy are negative and < -5 kcal/mol (Supplementary Table S3). Moreover, the proteins with the most compact binding affinity for scutellarin, 3,4-O-dicaffeoylquinic acid, 3,5-O-dicaffeoylquinic acid, 4,5-O-dicaffeoylquinic acid, and erigoster B are glutamate receptor 1 (GRIA1), PDE10A, protein kinase C gamma type (PRKCG), carbonic anhydrase 3 (CAR3), and GRIA1, respectively (Supplementary Table S3). It suggested the potential for high interaction of cell signaling-related proteins PDE10A and PRKCG, along with the chemical synaptic transmission-related protein GRIA1, with DZXI compounds. The visualization of their possible binding sites is shown in Figure 6.