Abstract Gelsemium is a medicinal plant that has been used to treat various diseases, but it is also well-known for its high toxicity. Complex alkaloids are considered the main poisonous components in Gelsemium. However, the toxic mechanism of Gelsemium remains ambiguous. In this work, network pharmacology and experimental verification were combined to systematically explore the specific mechanism of Gelsemium toxicity. The alkaloid compounds and candidate targets of Gelsemium, as well as related targets of excitotoxicity, were collected from public databases. The crucial targets were determined by constructing a protein–protein interaction (PPI) network. Subsequently, Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) were used to explore the bioprocesses and signaling pathways involved in the excitotoxicity corresponding to alkaloids in Gelsemium. Then, the binding affinity between the main poisonous alkaloids and key targets was verified by molecular docking. Finally, animal experiments were conducted to further evaluate the potential mechanisms of Gelsemium toxicity. A total of 85 alkaloids in Gelsemium associated with 214 excitotoxicity-related targets were predicted by network pharmacology. Functional analysis showed that the toxicity of Gelsemium was mainly related to the protein phosphorylation reaction and plasma membrane function. There were also 164 pathways involved in the toxic mechanism, such as the calcium signaling pathway and MAPK signaling pathway. Molecular docking showed that alkaloids have high affinity with core targets, including MAPK3, SRC, MAPK1, NMDAR[2B] and NMDAR[2A]. In addition, the difference of binding affinity may be the basis of toxicity differences among different alkaloids. Humantenirine showed significant sex differences, and the LD[50] values of female and male mice were 0.071 mg·kg^−1 and 0.149 mg·kg^−1, respectively. Furthermore, we found that N-methyl-D-aspartic acid (NMDA), a specific NMDA receptor agonist, could significantly increase the survival rate of acute humantenirine-poisoned mice. The results also show that humantenirine could upregulate the phosphorylation level of MAPK3/1 and decrease ATP content and mitochondrial membrane potential in hippocampal tissue, while NMDA could rescue humantenirine-induced excitotoxicity by restoring the function of mitochondria. This study revealed the toxic components and potential toxic mechanism of Gelsemium. These findings provide a theoretical basis for further study of the toxic mechanism of Gelsemium and potential therapeutic strategies for Gelsemium poisoning. Keywords: humantenirine, excitotoxicity, NMDA, NMDAR, network pharmacology, Gelsemium 1. Introduction Gelsemium, a genus of the Loganiaceae family, comprises three species: the Asian Gelsemium elegans (Gardner and Chapm.) Benth. and two North American related species, Gelsemium sempervirens (L.) J.St.-Hil. and Gelsemium rankinii Small [[36]1]. Gelsemium elegans, as a traditional Chinese medicine, has been used to treat skin disorders, malignant tumors and pain for a long time [[37]2,[38]3]. Gelsemium sempervirens is used in homeopathy to treat anxiety, neuralgia, migraine and spastic diseases [[39]4,[40]5]. Due to their variety and high biological activity, alkaloids are considered to be the main active substances in Gelsemium. According to the characteristics of chemical structure, alkaloids can be divided into six types: the gelsedine type, gelsemine type, humantenine type, koumine type, sarpagine type and yohimbane type [[41]3,[42]5]. The toxicity of different alkaloids is greatly different, and gelsedine-type and humantenine-type alkaloids are the most toxic in Gelsemium. The LD[50] of these two types of alkaloids for intraperitoneal injection in mice are usually lower than 1 mg·kg^−1 [[43]6,[44]7]. There are many cases of Gelsemium poisoning, some of which have even led to death [[45]8,[46]9], which seriously limits its application. Mechanistic studies for Gelsemium toxicity are scarce at present. A few studies have reported that the toxicity of Gelsemium is closely related to gamma-aminobutyric acid receptor (GABAR) [[47]10,[48]11] and glycine receptors [[49]12]. Recently, a phosphoproteomics study revealed that an N-methyl-D-aspartic acid receptor (NMDAR)-mediated excitotoxicity signaling pathway is linked to the death of gelsenicine (one of the toxic alkaloids in Gelsemium) poisoning [[50]13]. Therefore, it is hypothesized that the toxicity of Gelsemium is associated with excitotoxicity, which is consistent with the typical symptoms of Gelsemium poisoning, including dyspnea and convulsions. However, most of the existing studies on the toxicity of Gelsemium are still imperfect and the specific reasons for the toxicity differences among different alkaloids are still unclear. Network pharmacology [[51]14], a burgeoning interdisciplinary subject, highlights comprehensive thinking, focuses on the interaction among drugs, targets and diseases, and takes advantage of various means and technologies, such as molecular docking [[52]15] and enrichment analysis [[53]16], to screen the active ingredients, explore the potential core targets, and reveal the mechanisms of drugs. Network pharmacology is widely used in research to reveal the molecular mechanism of various drugs and poisons [[54]17], and the reliability and accuracy of the technical methods used have been recognized by international standards. In the present study, we used the network pharmacology method to find the possible targets of Gelsemium toxicity, and used the molecular docking method to verify the binding affinity of Gelsemium alkaloids with these core targets. Finally, an acute poisoning model of humantenirine in mice was established to further reveal the potential toxic mechanism of Gelsemium. This study is expected to lay a theoretical basis for the development and utilization of Gelsemium. 2. Materials and Methods 2.1. Collection and Screening of Gelsemium Alkaloids The alkaloid compounds in Gelsemium were obtained by referring to the literature [[55]3,[56]5] and searching public databases, including the Traditional Chinese Medicines Integrated Database (TCMID, [57]http://119.3.41.228:8000/tcmid/, accessed on 20 July 2022) [[58]18] and the Traditional Chinese Medicine Database @ Taiwan (TCMTW, [59]http://tcm.cmu.edu.tw/zh-tw/, accessed on 20 July 2022) [[60]19]. According to the Drug likeness parameters of the SwissADME platform([61]http://www.swissadme.ch/index.php, accessed on 22 July 2022) [[62]20], the alkaloids that have bioavailability scores ≥ 30% and meet at least two of the Lipinski rules (Lipinski, Ghost, Veber, Egan, and Muegge) were screened. The qualified alkaloids were finally determined to be candidate alkaloid components in Gelsemium. 2.2. Screening of Targets Corresponding to the Alkaloid Components The canonical SMILES of alkaloid components were retrieved from the open database PubChem ([63]https://pubchem.ncbi.nlm.nih.gov, accessed on 25 July 2022). Canonical SMILES were imported into the SwissTargetPrediction webtool ([64]http://www.swisstargetprediction.ch, accessed on 25 July 2022) [[65]21] to estimate the targets corresponding to each alkaloid in Homo sapiens. The targets with probability ≥0.1 were selected as potential targets. For the components not included in the SwissTargetPrediction database, the BATMAN-TCM online analysis tool ([66]http://bionet.ncpsb.org.cn/batman-tcm/index.php/Home/Index/index, accessed on 25 July 2022) was used as a supplement. The potential targets of alkaloids were obtained by taking a score cutoff of ≥ 10 and P ≤ 0.05 as screening conditions. The components without target information in both databases were excluded. The targets of all compounds were combined, and then the repeated targets were removed to obtain the targets corresponding to the alkaloid components of Gelsemium. 2.3. Identification of Targets Related to Excitotoxicity Targets for “excitotoxicity” were obtained from the GeneCards database ([67]https://www.genecards.org/, accessed on 28 July 2022) [[68]22] and National Center for Biotechnology Information databases (NCBI, [69]https://www.ncbi.nlm.nih.gov/, accessed on 28 July 2022). The obtained targets were summarized, and then the repetitive targets were eliminated to acquire the targets related to excitotoxicity. 2.4. Prediction of Targets of Alkaloid Components Associated with Excitotoxicity The intersection between the targets related to the alkaloid components of Gelsemium and excitotoxicity-associated targets was visualized by Venny 2.1 ([70]https://bioinfogp.cnb.csic.es/tools/venny/index.html, accessed on 02 August 2022) [[71]23]. 2.5. Protein-Protein Interaction (PPI) Network Construction and Analysis The targets of intersection were submitted to the STRING database ([72]https://string-db.org/cgi/input.pl, accessed on 02 August 2022) [[73]24] to construct the PPI network. The interaction score was set to 0.7, which indicates high confidence. In addition, the species was restricted to “Homo sapiens”. The result was saved as a “tsv” file. Finally, the results were input into Cytoscape 3.6.0 software to analyze core targets according to the Degree, ClosenessCentrality and BetweennessCentrality, which were used to evaluate the topological importance of nodes in the network [[74]25]. 2.6. Gene Ontology (GO) and Kyoto Encyclopedia Genes Genomes (KEGG) Pathway Enrichment Analysis GO and KEGG pathway enrichment analyses were carried out to explore the bioprocesses and signaling pathways involved in the excitotoxicity corresponding to alkaloids in Gelsemium. These targets were input into the Database for Annotation, Visualization and Integrated Discovery (DAVID, [75]https://david.ncifcrf.gov/, accessed on 02 August 2022) [[76]26]. Then, the results of the enriched GO terms, including biological process (BP), cellular component (CC) and molecular function (MF) were visualized by the bioinformatics ([77]http://www.bioinformatics.com.cn, accessed on 02 August 2022), as well as the dot bubble chart of KEGG pathway enrichment. 2.7. Construction of the Alkaloid–Target–Pathway Network A compound–target network and a target–pathway network were constructed by using Cytoscape v3.6.0 software. In the network, different types of nodes represent alkaloid components, targets and pathways. The connection between nodes indicates the interactions between components and targets or between targets and pathways. Three topological characteristic parameters (Degree, ClosenessCentrality and BetweennessCentrality) were used to identify the main poisonous components in Gelsemium. 2.8. Molecular Docking The crystal structures of MAPK3, SRC, MAPK1, NMDAR[2B] and NMDAR[2A] were obtained from the RCSB Protein Data Bank ([78]https://www.rcsb.org/, accessed on 18 August 2022, PDB codes: 4QTB, 2H8H, 4QTA, 7EU8 and 7EU7). All original ligands (including (3R)-1-(2-oxo-2-{4-[4-(pyrimidin-2-yl)phenyl]piperazin-1-yl}ethyl)-N-[3 -(pyridin-4-yl)-2H-indazol-5-yl]pyrrolidine-3-carboxamide, N-(5-CHLORO-1,3-BENZODIOXOL-4-YL)-7-[2-(4-METHYLPIPERAZIN-1-YL)ETHOXY]- 5-(TETRAHYDRO-2H-PYRAN-4-YLOXY)QUINAZOLIN-4-AMINE, (3R)-1-(2-oxo-2-{4-[4-(pyrimidin-2-yl)phenyl]piperazin-1-yl}ethyl)-N-[3 -(pyridin-4-yl)-2H-indazol-5-yl]pyrrolidine-3-carboxamide, and S-ketamine) and water molecules were removed, and hydrogen atoms and charges were added to the macromolecules. The three-dimensional structures of Gelsemium alkaloids downloaded from the PubChem database and optimized by Chem 3D Pro15.0 were used as the ligand. Molecular docking was finalized by AutoDock Vina [[79]27]. The size of the gridbox was fixed to 40 × 40 × 40 angstroms, with a spacing of 0.375 angstrom. All the parameters of the genetic algorithm (GA) were set to the default values. The conformers with the lowest binding energy were selected for analysis. 2.9. Animal Experiments 2.9.1. The LD[50] of Acute Humantenirine Poisoning Humantenirine was obtained from Chengdu Man Si Te Biotechnology Co., Ltd. (Chengdu, China) with a batch number of MUST-21052807 and purity of 98.4%. ICR mice (18–22 g) were provided by Hunan SJA Laboratory Animal Co., Ltd. (Changsha, China). The mice were reared in a standard facility. The animal experiments were approved by the Ethics Committee of Hunan Agricultural University (batch number 2020–43). The mice were randomly divided into 5 female groups and 5 male groups (n = 5). Humantenirine was injected into female mice intraperitoneally at 0.045, 0.056, 0.069, 0.086 and 0.11 mg·kg^−1. Male mice were given humantenirine by intraperitoneal injection at doses of 0.1, 0.12, 0.13, 0.16 and 0.18 mg·kg^−1. After administration, the poisoning symptoms and mortality were observed for 14 consecutive days. Finally, the lethal dose (LD[50]) was assessed by the Bliss method. 2.9.2. The Antidotic Effect of NMDA on Humantenirine Poisoning A total of 20 female mice were randomly assigned to two groups: the control group and the NMDA group (n = 10). In the control group, humantenirine, at a dose of 0.11 mg·kg^−1 according to the 100% lethal dose we explored earlier, was injected intraperitoneally into mice. The mice in the NMDA group were injected with 25 mg·kg^−1 NMDA intraperitoneally 20 min before humantenirine injection. The dose of NMDA used in the experiment was based on our previous exploration. Then, the death of the mice in the two groups was recorded. 2.9.3. Drug Treatment and Sample Collection Female mice were randomly assigned to three groups: (A) the control group, (B) the humantenirine group and (C) the NMDA group. The mice in group A were intraperitoneally injected with a certain volume of normal saline and then sacrificed. The brain tissue was removed on ice. The mice in group B received an intraperitoneal injection of humantenirine (0.11 mg·kg^−1). The brain tissues of poisoned dead mice were collected. NMDA (25 mg·kg^−1) was initially administered to the mice in group C, and humantenirine was given after 20 min. After the mice recovered, they were decapitated, and brain tissue was collected. The hippocampal tissue was separated on ice from the brain tissue of three randomly selected mice in each group for protein expression determination, while the hippocampal tissue of the other mice was prepared for the detection of ATP content and mitochondrial membrane potential. Measurement of ATP Content in the Hippocampal Tissue of Mice The hippocampal tissue samples were adequately homogenized with cold ATP extract solution after weighing. The homogenate was centrifuged at 8000× g/min at 4 °C for 10 min. Then, 0.5 mL chloroform was added to the supernatant, and the well-mixed solution was centrifuged at 10,000× g at 4 °C for 3 min. The supernatant was collected and used for the detection of ATP content according to the ATP Assay Kit. Determination of Mitochondrial Membrane Potential The hippocampal tissues were weighed and homogenized with precooled lysis buffer (1:10, w/v). After centrifugation at 1000× g/min at 4 °C for 5 min, the supernatant was collected and centrifuged again at 1000× g/min at 4 °C for 5 min. Then, the supernatant was transferred to another 2 mL microfuge tube and centrifuged at 12,000× g/min for 10 min. Next, 0.5 mL wash buffer was added to mitochondrial pellets to resuspend, and then centrifuged at 4 °C and 1000× g for 5 min. Finally, the supernatant was centrifuged at 12,000× g/min for 10 min. The obtained mitochondrial pellets were suspended in store buffer. The mitochondrial membrane potential was determined by using a JC-1 fluorescent probe. The mitochondrial suspension (20 μL) was added to 180 μL of JC-1 staining working solution (diluted 5 times with JC-1 staining buffer solution). The fluorescence intensity was detected by a fluorescence microplate reader. The excitation wavelength and emission wavelength of J-aggregates (red) were set to 525 nm and 590 nm, respectively. The excitation wavelength and emission wavelength of the JC-1 monomer (green) were set to 490 nm and 530 nm, respectively. The relative ratio of red/green fluorescence intensity was calculated to measure the proportion of mitochondrial depolarization. Detection of the Expression of Key Protein in Mice Hippocampus by Western Blotting The collected hippocampus samples were weighed and homogenized with RIPA lysis Buffer (Solarbio, China) containing phosphatase inhibitor (Coolaber, China) at low temperature, then lysed in an ice bath for 30 min. Next, the lysate was centrifuged at 13000 rpm at 4 °C for 10 min, and the supernatant was kept. The protein concentration was measured by BCA protein assay kit (CWBIO, China), and the protein samples were stored at −80 °C until use. The protein samples were separated by SDS-PAGE and transferred to PVDF membranes. Then, the membrane was blocked with protein-free rapid blocking buffer on a shaker for 10 min and washed with TBST 3 times, each time for 8 min. The membrane was incubated with the primary antibody (p44/42 MAPK (Erk1/2) (137F5) Rabbit mAb, 1:1000, Cell Signaling; Phospho-p44/42 MAPK (Erk 1/2) (Thr202/Tyr204) Rabbit mAb, 1:1000, Cell Signaling; β-Tubulin rabbit pAb, 1:4000, Proteintech) overnight at 4 °C. After washing with TBST, the appropriate secondary antibody (HRP-labeled goat anti-rabbit IgG (H+L), 1:5000, Biodragon) was added to incubate at room temperature for 1 h. After washing with TBST again, the BLT GelView 6000 Pro imaging system was used to visualize the protein bands. 3. Results 3.1. Putative Targets of Gelsemium Alkaloids Associated with Excitotoxicity The 98 alkaloid components that constitute Gelsemium were determined by references and databases, and 94 candidate alkaloids were obtained