Abstract Gelsemium elegans (Gardner and Champ.) Benth. (Gelsemiaceae) (GEB) is a toxic plant indigenous to Southeast Asia especially China, and has long been used as Chinese folk medicine for the treatment of various types of pain, including neuropathic pain (NPP). Nevertheless, limited data are available on the understanding of the interactions between ingredients-targets-pathways. The present study integrated network pharmacology and experimental evidence to decipher molecular mechanisms of GEB against NPP. The candidate ingredients of GEB were collected from the published literature and online databases. Potentially active targets of GEB were predicted using the SwissTargetPrediction database. NPP-associated targets were retrieved from GeneCards, Therapeutic Target database, and DrugBank. Then the protein-protein interaction network was constructed. The DAVID database was applied to Gene Ontology and Kyoto Encyclopedia of Genes and Genome pathway enrichment analysis. Molecular docking was employed to validate the interaction between ingredients and targets. Subsequently, a 50 ns molecular dynamics simulation was performed to analyze the conformational stability of the protein-ligand complex. Furthermore, the potential anti-NPP mechanisms of GEB were evaluated in the rat chronic constriction injury model. A total of 47 alkaloids and 52 core targets were successfully identified for GEB in the treatment of NPP. Functional enrichment analysis showed that GEB was mainly involved in phosphorylation reactions and nitric oxide synthesis processes. It also participated in 73 pathways in the pathogenesis of NPP, including the neuroactive ligand-receptor interaction signaling pathway, calcium signaling pathway, and MAPK signaling pathway. Interestingly, 11-Hydroxyrankinidin well matched the active pockets of crucial targets, such as EGFR, JAK1, and AKT1. The 11-hydroxyrankinidin-EGFR complex was stable throughout the entire molecular dynamics simulation. Besides, the expression of EGFR and JAK1 could be regulated by koumine to achieve the anti-NPP action. These findings revealed the complex network relationship of GEB in the “multi-ingredient, multi-target, multi-pathway” mode, and explained the synergistic regulatory effect of each complex ingredient of GEB based on the holistic view of traditional Chinese medicine. The present study would provide a scientific approach and strategy for further studies of GEB in the treatment of NPP in the future. Keywords: Gelsemium elegans (Gardner and Champ.) Benth, neuropathic pain, network pharmacology, molecular docking, molecular dynamics simulation Introduction Chronic pain condition is a major health issue that comprises five of the 11 top-ranking conditions lived with disability and is responsible for economic burden worldwide ([44]Vos et al., 2012; [45]Andrew et al., 2014). The prevalence of neuropathic pain (NPP) as a feature of chronic pain was estimated to range from 1 to 17.9% ([46]van Hecke et al., 2014). NPP is defined as an injury or disease of the somatosensory system involving complex pathogenesis according to the 2011 International Association for the Study of Pain ([47]Jensen et al., 2011). Overall, the current pharmacological interventions in NPP primarily consist of antidepressants or antiepileptics as the first-line treatments ([48]Lunn et al., 2014; [49]Moore and Gaines, 2019), lidocaine plasters, capsaicin high concentration patches, and tramadol as the second-line treatments ([50]van Nooten et al., 2017; [51]Kim et al., 2018), and strong opioids and botulinum toxin A as the third-line treatments ([52]Sommer et al., 2020). Unfortunately, patients with NPP conventional have an inadequate response with only 40–60% of patients achieving partial relief to the current pharmacological therapy and suffering from side effects include sedation, anticholinergic effects, nausea, and orthostatic hypotension ([53]Dworkin et al., 2007; [54]Cavalli et al., 2019). Therefore, there is a necessity to explore more effective analgesics with novel mechanisms and low side effects for the treatment of NPP. Traditional Chinese medicine (TCM) is an abundant resource for drug development and provides innovative insight into therapeutic approaches. Gelsemium elegans (Gardner and Champ.) Benth. (Gelsemiaceae) (GEB) is a toxic plant indigenous to Southeast Asia especially China, which has long been used as Chinese folk medicine for the treatment of various types of pain, such as neuralgia, sciatica, rheumatoid arthritis, and acute pain ([55]Rujjanawate et al., 2003; [56]Lin et al., 2021). Phytochemical studies have revealed that the main active ingredients of GEB are alkaloids, especially the indole alkaloids, such as koumine, gelsemine, gelsenicine, and gelsevirine ([57]Jin et al., 2014). These alkaloids are distributed throughout the whole plant, especially rich in the roots. GEB and its active alkaloids have been studied increasingly and exert promising pharmacological effects in NPP. It was reported that a crude alkaloidal extract solution from GEB could significantly increase the pain thresholds of mice in both hot plate and writhing tests at the dose of 0.5, 1.0, and 2.0 mg/kg ([58]Rujjanawate et al., 2003). As an important active ingredient, previous studies indicated that koumine exhibited a significant analgesic effect in vitro and in several animal models of NPP. These studies suggested that koumine alleviated NPP may through a wide variety of mechanisms, including enhancing 3α-hydroxysteroid oxidoreductase mRNA expression and bioactivity ([59]Qiu et al., 2015) in the spinal cord, upregulating allopregnanolone ([60]Xu et al., 2012), and inhibiting astrocyte activation as well as M1 polarization while sparing the anti-inflammatory responses to NPP ([61]Jin et al., 2018a; [62]Jin et al., 2018b). Other active ingredients, gelsemine, gelsenicine, and gelsevirine may produce antinociception by activating the spinal α3 glycine/allopregnanolone pathway ([63]Zhang and Wang, 2015). However, all the existing studies focused on limited ingredients, targets, and pathways, and lacked the integral thoughts and exploration on TCM with multiple ingredients and targets. Hence, the interactions between ingredients-targets-pathways and other underlying molecular mechanisms of GEB against NPP remain unclear. Network pharmacology is mostly used to screen the active ingredients, predict the corresponding target, and explore the comprehensive molecular mechanisms of TCM. The key ideas of network pharmacology are based on the theory of system biology and multi-direction pharmacology, which are consistent with the holistic philosophy of TCM ([64]Li and Zhang, 2013). Molecular docking simulation is a computational method for exploring the ligand conformations adopted within the binding sites of receptors in the intermolecular recognition process ([65]Ferreira et al., 2015). Different from traditional pharmacological research methods of TCM, network pharmacology-based analysis combined with molecular docking technology could provide a new perspective for the study of the molecular mechanism of TCM. In the present study, we proposed an “ingredient-target-pathway” network to reveal the potential material basis and compatibility molecular mechanisms of GEB against NPP based on the network pharmacology and experimental evidence. The flowchart of our work is shown in [66]Figure 1. FIGURE 1. [67]FIGURE 1 [68]Open in a new tab The flowchart of network pharmacology analysis. Materials and Methods Identification of Active Ingredients in GEB The potential active ingredients in GEB were retrieved from the published literature ([69]Jin et al., 2014) and the online public databases, including the Traditional Chinese Medicines Integrated database (TCMID) ([70]http://www.megabionet.org/tcmid/) ([71]Huang et al., 2018), Bioinformatics Analysis Tool for Molecular mechanism of Traditional Chinese Medicine (BATMAN-TCM) ([72]http://bionet.ncpsb.org/batman-tcm/) ([73]Liu et al., 2016), and Traditional Chinese Medicine database@ Taiwan ([74]http://tcm.cmu.edu.tw/zh-tw/) ([75]Chen, 2011). Active ingredients with oral bioavailability (OB) ≥ 30% and drug-likeness (DL) ≥ two of five features (Lipinski, Ghose, Veber, Egan, and Muegge) were selected, which was recommended by SwissADME ([76]http://www.swissadme.chwebsite) ([77]Daina et al., 2017). The final cluster of chemical ingredients of GEB was determined after removing duplicates. Identification of Ingredients-Related Targets Targets of the active ingredients were predicted using SwissTargetPrediction ([78]http://www.swisstargetprediction.ch), a popular online server that could accurately predict the targets of bioactive molecules with known ligands ([79]Gfeller et al., 2014). 3D structural SDF formats (.sdf) of active ingredients of GEB were acquired from the PubChem database ([80]https://pubchem.ncbi.nlm.nih.gov/) and imported into SwissTargetPrediction for identification of potential drug targets in humans. After removing duplicate targets, the targets of ingredients with SwissTargetPrediction probability ≥0.1 were chosen as potential targets, and compounds without target information were excluded. Identification of Disease-Associated Targets The disease-associated targets of NPP were collected from GeneCards ([81]https://www.genecards.org/) ([82]Safran et al., 2010), the Therapeutic Target database ([83]https://db.idrblab.org/ttd/) ([84]Wang et al., 2020a), and DrugBank ([85]https://go.drugbank.com) ([86]Wishart et al., 2018). “Neuralgias”, “Neuropathic Pain”, “Neurodynia”, and “Nerve Pain” were used as keywords in the three databases and Homo sapiens targets with a disease relevance score ≥ of three were selected for the study. Topology Analysis of the Protein-Protein Interaction (PPI) Network The intersection of ingredients-related targets and disease-associated targets was visualized by overlapping with a Venn diagram. Then, a PPI network was constructed through the String database ([87]https://stringdb.org/) to explore the core regulatory genes ([88]Szklarczyk et al., 2019). PPI information was extracted with an interaction score of 0.4 and the species was only limited to “Homo sapiens”. The topology analysis of the PPI was performed with Cytoscape 3.7.2 ([89]http://cytoscape.org/.ver.3.7.2). NetworkAnalyzer analysis was used to screen key targets according to the degree value. The top 15 important proteins with a higher level of degrees in the interaction network were considered as the key targets for GEB in the treatment for NPP. Furthermore, the Molecular Complex Detection (MCODE) plugin was used to detect cluster modules from the complex network with the node score cutoff of 0.2, K-core of 2, and degree cutoff of 2. Gene Ontology (GO) and Kyoto Encyclopedia Genes Genomes (KEGG) Enrichment Analysis The GO and KEGG enrichment analysis were performed to explore the signaling pathways and bioprocesses involved in the key targets. The database for Annotation, Visualization, and Integrated Discovery (DAVID, [90]https://david.ncifcrf.gov/.ver.6.8) was applied to conduct the enrichment analysis ([91]Dennis et al., 2003). The species was limited to “Homo sapiens”, and the enrichment of pathway was considered significant when the modified fisher exact false discovery rate (FDR) < 0.01. The results of the KEGG pathway and enriched GO terms of biological processes (BP), cell composition (CC), and molecular function (MF) were visualized by the R software package (3.5.2). Construction of “Ingredient-Target-Pathway” Network The “ingredient-target-pathway” networks including the potential ingredients-targets network of GEB against NPP and targets-pathways network of GEB against NPP were constructed by Cytoscape ([92]Shannon et al., 2003). In the network, nodes represent the final active ingredients and targets, while the connections between the nodes represent the interactions between these biological processes and signaling pathways. Three key topological parameters were used to evaluate the topological coefficients between nodes: “degree" (the number of connections between the molecular and target in the core architecture of the network), “betweenness” (the number of shortest paths of a node to the total number of paths through all nodes), and “closeness” (the inverse of the sum of the shortest paths from a node to other nodes in the network). Ingredients-Targets Molecular Docking Molecular docking was used to predict the interactions between core active ingredients of GEB and proteins selected from the center targets from a molecular perspective. 3D structures of active ingredients in SDF (.sdf) format were selected from the PubChem database ([93]https://pubchem.ncbi.nlm), and the crystal structures of the target proteins were downloaded from the PDB database ([94]https://www.rcsb.org/) with a crystal resolution of less than 2 Å. Molecular docking was performed by importing the crystal structure into the Pymol 2.4.1 Software ([95]https://pymol.org/2/) for dehydration, hydrogenation, and ligand separation. Thereafter, Autodock Vina 1.1.2 software was used to construct a crystal structure docking grid box for each target. Then the molecules with the lowest binding energy for each active compound in the docking conformation were allowed for semi-flexible docking by comparing with the original ligands and intermolecular interactions (hydrophobicity, cation-π, anion-π, π-π stacking, hydrogen bonding, etc.). Box center coordinates and size of the box were determined for evaluating the interaction. The results were analyzed and visualized using Pymol, and the numbers of grid points in the three dimensions used in this study were 40 40 40 0.375. Molecular Dynamics Simulation of Ligand Complex The molecular dynamics simulation study is employed to assess the stability and interaction between the protein and ligands after docking. The simulation run was performed for 100 ns using the NVIDIA RTX 1060 GPU accelerated GROMACS 2020 software molecular dynamics package. In the preliminary stage, the Charmm36 force field was used for the protein parameters. The CGenFF server was used for the ligand topology, and a TIP3P water model with appropriate Na^+/Cl^− ions was subsequently generated and neutralized the charge of the system. The system converged to a minimum energy level using the steepest descent method of 50,000 steps and <10.0 kJ/mol force. Then, the equilibration process was conducted with 100 ps for constant NVT (number, volume, and temperature) heating to 300 K, followed by 100 ps for constant NPT (number of particles, pressure, and temperature) with a time step of 2 fs. The bonds of atoms were restrained by recruiting the LINCS algorithm. After the processes of energy minimization and equilibration, the molecular dynamics simulation was conducted the leap-frog algorithm for 100 ns with a time step of 2 fs. The geometrical parameters of the systems, such as root mean square deviation (RMSD) and root mean square fluctuation (RMSF), were determined and compared with the primitive ligand complex. Experimental Verification in Chronic Constriction Injury (CCI) Rat Model Koumine (99% purity) was isolated from GEB as described by Su et al. ([96]Su et al., 2011), and it was dissolved or diluted in sterile physiological saline (0.9% w/v NaCl). Male Sprague–Dawley rats (180–200 g) were purchased from Shanghai Laboratory Animal Center, Chinese Academy of Sciences. The rats were adapted in the condition of 25 ± 2°C with a 12-h light/dark cycle (lights on at 8:00 am) and free access to standard laboratory food and water. The experiments met the requirements of guidelines for animal care and the use of Fujian Medical University. The experimental protocols were reviewed and approved by the Committee of Ethics of the Fujian Medical University (Fujian, China). Animals were assigned into different groups: the sham control group (rats underwent the surgical procedures without any manipulation related to nerve injury), the CCI model group (rats received the vehicle, 0.28, 1.4, 7.0 mg/kg of koumine). The dose used in the experimental assay was based on the published literature ([97]Jin et al., 2018b), and no adverse effects and sedative effects were observed in the rats. The CCI rat model was performed according to the method described by Bennett et al. ([98]Bennett and Xie, 1988). Behavior tests consist of thermal hyperalgesia and mechanical allodynia tests. The thermal hyperalgesia test using a commercial thermal paw stimulator (PL-200, Chengdu Technology and Market Co., Ltd., Sichuan, China) was evaluated before operation (baseline), drug administration (pre-dosing), and 6, 8, 10, 12, and 14-days after drug administration (post-dosing), and paw thermal withdrawal latency (TWL) was calculated as described by Hargreaves et al. ([99]Hargreaves et al., 1988). The mechanical allodynia test was measured with a commercially available electronic von Frey apparatus (Model 2390; IITC Life Science Inc., Woodland Hills, CA), and each hind paw and mechanical withdrawal latency (MWL) was calculated 30 min after the TWL measurement according to the published literature ([100]Mitrirattanakul et al., 2006). The observer measuring the behaviors was blind to drug pretreatments in all behavioral tests. Then, rats were anesthetized by chloral hydrate, and the lumber segments (L5-L6) of the spinal cord were dissected, weighed, and stored at −80°C. Then, the lumber segments were homogenized for 30 min in an ice bath with RIPA lysis buffer (CoWin Biosciences, China) containing phosphatase inhibitor (CoWin Biosciences, China) after ultrasonic crushing. Protein concentrations were determined using an enhanced BCA protein assay kit (Beyotime Biotech Inc., China), and the protein samples were stored at −80°C until use. Total protein samples were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred onto the nitrocellulose membrane. The membrane was blocked with 5% non-fat dried milk in tris buffer for 1 h at room temperature and washed with tris-buffered saline and tween 20 every 10 minutes for three times. Then, the membrane was incubated with antibodies (EGFR rabbit pAb: A11577, 1:500, ABclonal; JAK1 rabbit pAb: A5534, 1:500, ABclonal; AKT1 rabbit mAb: A17909, 1:500, ABclonal; β-actin rabbit mAb: AF1186, 1:1,000, Beyotime) overnight at 4°C. After incubation with the appropriate secondary antibodies (HRP-labeled goat anti-rabbit IgG: A0208, 1:1,000, Beyotime) at room temperature for 1 h, the protein blots were visualized in the ChemiDoc XRS imaging system (Bio-Rad, CA). Results Putative Targets of GEB Against NPP A total of 98 compounds in GEB were retrieved from published literature and online databases, and 57 potentially active ingredients were filtered by OB and DL provided by SwissADME ([101]Table 1). 679 potential targets were eventually predicted based on the SwissTargetPrediction after eliminating duplicate targets ([102]Supplementary Table S1). 1,047 targets related to NPP were obtained through the Gene Cards database. Out of these targets, 367 potential targets were finally screened out with a disease relevance score ≥3 ([103]Supplementary Table S2). Subsequently, as shown in [104]Figure 2, 679 GEB ingredients-related targets were intersected with 367 NPP disease-related target genes using Venn diagrams to identify 52 putative targets between GEB and NPP, which were considered candidate targets of GEB against NPP. TABLE 1. Information of the active compounds in GEB for network analysis. NO. Name Compound CID MW MF Source 1 N-methoxyanhydrovobasinediol [105]102004539 338.4 C[21]H[26]N[2]O[2] References