Information regarding the specific plants in CH was extracted from the Traditional Chinese Medicine System Pharmacology Database (TCMSP) (9). The molecular structures of every compound were obtained from the NCBI PubChem database (10) and confirmed through thorough research in literature. It is crucial to evaluate the absorption, distribution, metabolism, and excretion characteristics of every compound in order to identify the active components of CH and examine their functional foundation. For this investigation, we opted for oral absorption (OA) and drug similarity (DS). For the purpose of further analysis, molecules with OB ≥ 30% were chosen in this particular section (11). In TCM research, the DL threshold of 0.18, which is based on the average value of all compounds in the DrugBank database (12), is widely used as a screening standard for identifying ‘drug-like’ compounds.

In this process, the active ingredients filtered out early were used as bait to identify the corresponding targets in the TCMSP and DrugBank databases. Afterwards, the UniProt KB (13) was used to map all stationary objectives and acquire their official gene symbols.

The Gene Cards database (14), OMIM database (15), PharmGkb database (16), TTD (17, 18), and DrugBank database were used to screen AP-related targets. We combined the targets from these five databases, removed duplicate targets, and identified the disease targets related to AP. In this section, the search was conducted using the keywords “acute pancreatitis” and selection of species condition for “homo sapiens” as a supplement. Subsequently, the retrieved targets were mapped to UniProt KB to obtain the corresponding official gene symbols.

In the end, we compared the projected goals of CH active components with the targets related to AP and chose the common targets as potential therapeutic targets for CH. A Venn diagram of the targets was plotted and visualized using FunRich software in the CH-AP network. Subsequently, the common targets were processed for additional analysis.

In order to delve deeper into the possible molecular mechanism of CH, we conducted GO enrichment analysis and KEGG signaling pathway analysis. This involved connecting the chosen possible targets to the Database Visualization and Integrated Discovery system (DAVID) Bioinformatics Resources 6.8 (19). Only the terms with P < 0.05 significance level were taken into account in this particular section. The overlapping targets were plotted using R language, and the GO results were represented using a bar chart, while the KEGG analysis results were represented using a bubble chart.

Drug-compound-target (D-C-T) network of CH was created by establishing connections between active compounds and their potential targets using Cytoscape v3.9.1 software (20), a robust tool frequently employed in bioinformatics research for data integration and visualization. Nodes in graphical networks depict compounds and targets, while edges represent depict interactions between the compound and target. Moreover, ‘degree’, a crucial topological parameter in network pharmacology, was examined using the CytoNCA plugin of Cytoscape. The degree of a node indicates the total number of other nodes that interact with it.

Construction of a network for the interaction between proteins (PPI) was carried out to identify potential therapeutic targets of the active components in CH. The STRING database (21) was utilized for this purpose. The settings for network analysis consisted of choosing the mode of analysis as ‘Multiple proteins’, selecting ‘Homo sapiens’ as the species, and setting a confidence score of at least 0.95. Isolated proteins were excluded, and the results were saved as a TSV file. Afterwards, we employed the Cytoscape 3.9.1 application to exhibit the protein-protein interaction (PPI) network. To pinpoint the main targets, we carried out network topology analysis through the “Network Analyzer” module in Cytoscape. Different parameters, including Betweenness (BC), Closeness (CC), Degree (DC), Eigenvector (EC), local average connectivity-based methods (LAC), Network (NC), and Information (IC), were used to establish the main objectives. During this process, the core targets were identified by assessing the degree of each node in the network.

After conducting the KEGG pathway analysis on the potential targets identified in the DAVID database, the Compound-Target-Pathway (C-T-P) network was built in Cytoscape 3.9.1. This network connected the top 30 KEGG signaling pathways with their respective targets. Nodes in graphical networks represent compounds, targets, and signaling pathways, while edges represent C-T-P interactions. A node’s significance in the network is indicated by a higher degree value.

The network known as the ‘Compound-Target-Pathway (C-T-P)’ was employed to discover compounds and their possible targets. These targets were subsequently analyzed through molecular docking investigations with AutoDock Vina (22). Molecular docking involves several steps. Initially, the PDB database (23) was utilized to acquire the three-dimensional arrangement of the main protein of interest. Second, this structure was imported into PyMOL software to eliminate water molecules and inactive ligands. Following this, the structure was loaded into AutoDockTools software for hydrogenation and charging, and finally saved in PDB format. Third, the target proteins and the components of CH were transformed into the pdbqt file format, which is in line with AutoDock Vina’s compatibility. To facilitate the docking of the target protein’s active site, the size and coordinates of the Grid Box were adjusted. Subsequently, the active components of CH were docked onto the active site of the target protein. Finally, the components that had the best binding energy and strongest affinity towards the target proteins were selected. Using the PyMOL software, we visualized the docked complexes of the selected elements.

Each bag of CH contains 7 g. The recommended dosage for adults with AP is 42 g, three times per day. In the case of rats, the equivalent dosage was 6.3 times that in adults. Therefore, the dosage of CH in the rats was approximately 0.7 g/kg.BW × 6.3 ≈ 4.4 g/kg.BW = 0.44 g/100g.BW. CH was dissolved in saline at a 0.44 g/mL concentration. The rats were gavaged with 1 mL of the solution per 100 g of body weight.

Sodium taurocholate (Item No. The purity of 97-98% was acquired from Beijing Bioway Technology Co., Ltd. The CH ready-to-use granules were obtained from the Affiliated Traditional Chinese Medicine Hospital of Southwest Medical University. Ruixin Biotech was the source of the purchased ELISA kits (RX302856R, RX302869R, and RX302058R). An apoptosis detection kit (T-6013) was purchased from UElandy (China). Antibodies against p-PI3K (ab182651) were purchased from Abcam (Cambridge, USA). The β-actin antibody (AT0040) was purchased from Engibody (USA). Proteintech (USA) provided the antibodies for phosphoinositide 3-kinase (PI3K) (60225-1-Ig), Bcl-2 (68103-1-Ig), P-P65 (82335-1-RR), and P65 (10745-1-AP). Cell Signaling Technology (Danvers, MA, USA) was the source of the purchased antibodies targeting Bax (14796S), AKT (8200S), and p-Akt (8200S). All additional substances utilized in this research were of analytical quality and obtained from nearby vendors.

The pancreas tissue was preserved in a 4% solution of paraformaldehyde, underwent dehydration using a gradient method, was embedded in paraffin wax. Hematoxylin and eosin (HE) staining was performed on thick sections sliced to a 5 μm thickness. Pancreatic abnormalities following HE staining were evaluated by two pathologists using the Schmidt pathology scoring criteria (25) under a light microscope, employing a double-blind approach. Five randomly selected fields were assessed to determine the scores for edema, inflammation, and necrosis. Ultimately, the mean score was used to quantify the extent of damage.

The paraffin sections were deparaffinized in water. Proteinase K was used to permeabilize the tissues. The TUNEL reaction solution was applied to tissues and incubated in the dark. Finally, the sections were sealed after DAPI counterstaining. Fluorescence microscope was utilized to capture images and ImageJ software was employed for quantification.

Protein specimens were obtained by utilizing RIPA lysis solution comprising of protease and phosphatase inhibitors, and protein concentration was ascertained utilizing a BCA assay kit. Following the electrophoretic separation of proteins, PVDF membranes were utilized to facilitate their transfer. Next, the antibodies were encapsulated with 5% BSA for a duration of 30 minutes. Subsequently, the membrane was incubated with the specified antibodies at 4°C overnight. After undergoing three 10-minute washes with TBST, the membrane underwent incubation with a secondary antibody conjugated with horseradish peroxidase for a duration of 2 hours. Subsequently, the membrane was washed again and subjected to incubation with an ECL chemiluminescent substrate to facilitate imaging. The obtained images were analyzed using the ImageJ software.

ELISA was used to detect the serum levels of interleukin (IL)-6, IL-1β, and tumor necrosis factor-alpha (TNF-α) in accordance with the manufacturer’s guidelines. Additionally, an automated biochemical analyzer was utilized to measure the levels of serum amylase and lipase.

By screening the TCMSP database, a total of 168 active components and 276 targets were identified. The results indicate that most of the compounds interact with multiple targets ( Supplementary Table S1 ). The active components of Baishao, Chaihu, Chishao, Dahuang, Danshen, Gancao, Houpu, Huangqin, Huangqi, Taoren, Yanhusuo, Zhishi, and Zhizi were 3, 7, 8, 5, 28, 40, 2, 13, 7, 16, 42, 4, and 7, respectively. The 10 molecules with the highest OB scores are listed in Table 1 .

By eliminating duplicates, we acquired a grand total of 1986 disease targets from the GeneCards database using a relevance score of ≥10. Out of these, 2060 targets were obtained, including 143 targets from the OMIM database, 112 targets from the PharmGkb database, 9 targets from the TTD database, and 6 targets from the DrugBank database. We identified overlapping targets to select 177 potential therapeutic targets for treating AP between the compound and AP-related targets, as shown in Supplementary Table S2 . The corresponding Venn diagram is shown in Figure 2A .

DAVID’s GO analysis yielded three distinct categories: biological processes (BP), cellular constituents (CC), and molecular functions (MF), revealing a total of 4544, 269, and 439 identified terms in each category, respectively. The top 10 GO terms significantly enriched in each stratum can be found in Figure 2B . The results showed that BP was most related to epithelial cellular proliferation, CC was most related to the vesicle lumen, and MF was most related to mRNA blinding involved in post-transcriptional gene splicing.

In order to enhance our comprehension of the pharmacological mechanism of CH for AP treatment at the signaling pathway level, we conducted KEGG analysis. This analysis unveiled that the chosen 177 potential targets were significantly present in 186 pathways, encompassing the PI3K/AKT signaling pathway (hsa04151), AGE/AGE signaling pathway (hsa04933), JAK/TAT signaling pathway (hsa04630), and TNF signaling pathway (hsa04668). The top 30 most significantly enriched pathways are shown (P<0.05, ordered by gene count) in Figure 2C . A comprehensive overview of the PPI data is available in Supplementary Table S3 .

Cytoscape 3.9.1 was used to construct and visualize the D-C-T network of the active CH compounds and potential targets, which includes common targets between compound targets and AP-related targets. In this section, the 181 compounds linked to potential targets are presented in the network diagram. Figure 2D illustrates a network consisting of 358 nodes, including 181 compounds and 177 potential targets, connected by 1832 edges. The key compounds of CH can be considered as the ones that showed strong interactions in the C-T network. These compounds are from Baishao, Chaihu, Chishao, Dahuang, Danshen, Gancao, Houpu, Huangqin, Huangqi, Taoren, Yanhusuo, Zhishi, and Zhizi, with the numbers 3, 7, 8, 5, 28, 40, 2, 13, 7, 16, 42, 4, and 7, respectively. 148 different herb ingredients targeted prostaglandin G/H synthase 2 (PTGS2), while heat shock protein (HSP) 90-alpha (HSP90AA1), sodium channel protein type 5 subunit alpha (SCN5A), prostaglandin G/H synthase 1 (PTGS1), and nuclear receptor coactivator 2 (NCOA2) were targeted by 89, 83, 82, and 74 herb ingredients, respectively. This implies that the synergistic therapeutic effect on AP is achieved by multiple CH compounds.

The ingredients of CH are represented by the outer nodes in Figure 3 , while the common targets between the drug and the disease are represented by the middle nodes. The CH formula uses various colors to represent different herbal medicines: navy blue for Chaihu (Bupleuri Radix), crimson for Houpo (Magnoliae Officinalis Cortex), pale green for Chishao (Paeoniae Radix Rubra), vibrant green for Dahuang (Rhei Radix et Rhizoma), ochre yellow on the right for Taoren (Persicae Semen), dark green for Danshen (Salviae Miltiorrhizae Radix et Rhizoma), rose for Gancao (Glycyrrhizae Radix et Rhizoma), lavender for Yanhusuo (Corydalis Rhizoma), yellow on the left for Huangqi (Astragali Radix), golden yellow for Huangqin (Scutellariae Radix), deep violet for Zhishi (Aurantii Fructus Immaturus), pale yellow for Zhizi (Gardeniae Fructus), and sky blue for Baishao (Paeoniae Radix Alba).

Utilization of the STRING database, a PPI network of the chosen 177 potential targets was established based on the aforementioned methods according to the minimum required interaction score ≥ 0.95 with isolated nodes hidden in the network in Figure 4B . In CytoNCA, different measures such as betweenness (BC), closeness (CC), degree (DC), eigenvector (EC), local average-connectivity-based methods (LAC), networks (NC), and information (IC) were utilized for the scoring and filtering procedures. Each gene was required to surpass its median value for retention. This scoring and filtering process was performed thrice in Figure 4A . Ultimately, 14 core target genes were identified (STAT3, IL6, MYC, CDKN1A, AKT1, MAPK1, MAPK3, MAPK14, HSP90AA1, HIF1A, ESR1, TP53, FOS, and RELA) comprising 14 nodes and 60 edges. Detailed information on PPI data is presented in Supplementary Table S4 .

The 14 core targets identified in the PPI network were linked to their corresponding compounds in the DCT network through the top 30 KEGG pathways, as depicted in Figure 4C . The C-T-P network that was obtained consisted of 165 nodes, comprising 121 compounds, 14 targets with potential, and 30 pathways for signaling. Additionally, there were 485 edges in the network. Additional examination uncovered that the chosen possible objectives had a strong connection with the PI3K/AKT (hsa04151, degree = 15), TNF (hsa04668, degree = 12), IL-17 (hsa04657, degree = 9), and HIF-1 (hsa04066, degree = 9) signaling pathways. The involvement of these pathways is crucial in the advancement of AP and various other inflammatory conditions.

From the C-T-P network, we identified the top three selected compounds and their corresponding target proteins, and conducted a molecular docking validation. Specifically, the target proteins chosen for docking overlapped with key target proteins in the PPI network. Table 2 displays the outcomes of molecular docking. A lower docking score indicates a stronger ligand-receptor binding affinity, signifying a higher likelihood of interaction. All the binding energy scores between the active compounds and the key target proteins were less than -5.0 kJ/mol. The 3D binding models of each target protein and compound are shown in Figure 3 . Additionally, the docking diagrams of the compound-target proteins suggested possible intermolecular interactions between the core compounds and protein targets.

According to the findings shown in Figure 5A , the group with AP demonstrated significantly elevated serum amylase and lipase levels compared to the sham group. However, the administration of CH resulted in decreased levels. Moreover, the AP model group exhibited notably elevated levels of edema and inflammation compared to the group that underwent a sham operation. However, the administration of CH resulted in a decreased level of pancreatic pathological harm, which involved decreased swelling, inflammation, and tissue death, along with a decrease in pathological scores (P < 0.05) ( Figures 5B, C ). Collectively, these results demonstrate that CH alleviates pancreatic tissue inflammation and mitigates pathological damage in AP rats.

According to Figure 6 , the AP group showed significantly raised levels of serum TNF-α, IL-6, and IL-1β in comparison with the sham group. However, the administration of CH resulted in a decrease of these levels. Over time, the levels of TNF-α, IL-6, and IL-1β in the AP group exhibited a substantial growth (P < 0.05).

The levels of PI3K, AKT, and NF-ϰB proteins did not show any notable differences among the sham, AP, and CH groups ( Figure 7 ). However, the AP model group exhibited considerably elevated levels of protein phosphorylation compared to the sham group. Conversely, the intervention group CH demonstrated notably reduced levels of phosphorylation among the proteins PI3K, AKT, and NF-ϰB.

TUNEL staining was used to analyze apoptosis of acinar cells. Pancreatic acinar cells underwent apoptosis upon the induction of sodium taurocholate, as depicted in Figure 8A . After CH treatment, the incidence of TUNEL-positive cells has significantly increased, unlike the AP group ( Figure 8B ). Furthermore, the western blot analysis demonstrated that the CH group exhibited markedly elevated levels of the pro-apoptotic protein Bax, while displaying notably reduced levels of the anti-apoptotic protein BCL-2 in comparison to both the sham and AP groups. This indicates that CH significantly enhanced apoptosis in acute pancreatitis. ( Figures 8C–E )