Development of Novel Herbal Compound Formulations Targeting Neuroinflammation: Network Pharmacology, Molecular Docking, and Experimental Verification

Neuroinflammation plays an important role in the onset and progression of neurodegenerative diseases. The multicomponent and multitarget approach may provide a practical strategy to address the complex pathological mechanisms of neuroinflammation. This study aimed to develop synergistic herbal compound formulas to attenuate neuroinflammation using integrated network pharmacology, molecular docking, and experimental bioassays. Eight phytochemicals with anti-neuroinflammatory potential were selected in the present study. A compound-gene target-signaling pathway network was constructed to illustrate the mechanisms of action of each phytochemical and the interactions among them at the molecular level. Molecular docking was performed to verify the binding affinity of each phytochemical and its key gene targets. An experimental study was conducted to identify synergistic interactions among the eight phytochemicals, and the associated molecular mechanisms were examined by immunoblotting based on the findings from the network pharmacology analysis. Two paired combinations, andrographolide and 6-shogaol (AN-SG) (IC50 = 2.85 μg/mL), and baicalein-6-shogaol (BA-SG) (IC50 = 3.28 μg/mL), were found to synergistically (combination index <1) inhibit the lipopolysaccharides (LPS)-induced nitric oxide production in microglia N11 cells. Network pharmacology analysis suggested that MAPK14, MAPK8, and NOS3 were the top three relevant gene targets for the three phytochemicals, and molecular docking demonstrated strong binding affinities of the phytochemicals to their coded proteins. Immunoblotting suggested that the AN-SG and BA-SG both showed prominent effects in inhibiting inducible nitric oxide synthase (iNOS) (p < 0.01 and p < 0.05, respectively) and MAPKp-p38 (both p < 0.05) compared with those induced by the LPS stimulation only. The AN-SG combination exhibited greater inhibitions of the protein expressions of iNOS (p < 0.05 vs. individual components), which may partly explain the mechanisms of the synergy observed. This study established a practical approach to developing novel herbal-compound formulations using integrated network pharmacology analysis, molecular docking, and experimental bioassays. The study provides a scientific basis and new insight into the two synergistic combinations against neuroinflammation.


Introduction
Neurodegenerative diseases are characterised by the progressive loss of neurons and the deposition of proteins manifested as altered physicochemical properties in the brain and peripheral organs [1]. It is a group of major diseases in the elderly that signifcantly impacts families, communities, and healthcare systems worldwide. Te etiologic and underlying pathophysiology of neurodegenerative diseases is complex and mediated by various factors [2]. To date, there is no treatment available to prevent or cure neurodegenerative diseases due to the lack of understanding of the causes and pathological mechanisms [3][4][5].
Neuroinfammation refers to the activation of the brain's innate immune system and the abnormal secretion of proinfammatory cytokines in response to an infammatory challenge [6]. Emerging evidence has shown that neuroinfammation is associated with the onset and development of many neurodegenerative diseases, such as Alzheimer's disease, Parkinson's disease, and amyotrophic lateral sclerosis [7]. Microglia are the resident immune cells in the central nervous system that play a crucial role in normal brain function and neuroinfammation-mediated neuronal pathology [8]. Under the pathological conditions, the excessive activation of microglia triggers the elevated production of proinfammatory mediators [e.g., nitric oxide (NO) and tumour necrosis factor (TNF)-α] that lead to the high permeability of the blood-brain barrier (BBB) and impaired neuronal survival [9]. Tereby, neuroinfammation has been implicated as an important therapeutic target for neurodegenerative diseases [10,11].
Tere has been increasing interest in studying phytochemicals for their potential anti-neuroinfammatory efects [12,13]. In the current study, a literature review was conducted to identify phytochemicals that possess anti-neuroinfammatory potential, and those that are commercially available were sourced for preliminary laboratory testing. Based on the results [14], eight phytochemicals were fnally selected in the present study. Luteolin (LU), a favonoid found in various vegetables, medicinal herbs and fruits, possesses anti-infammatory, antioxidant, neuroprotective, neurotrophic, and neurogenesis activities [15]. Baicalein (BA), a main bioactive ingredient from the root of Scutellaria baicalensis Georgi, exhibits potent anti-neuroinfammatory and neuroprotective efects [16]. Andrographolide (AN) is a major active constituent of Andrographis paniculate Burm. f., that has been shown to have potent antioxidant and antineuroinfammatory properties [17,18]. 6-shogaol (6-SG), a bioactive ingredient in dried ginger, possesses a strong anti-neuroinfammatory property [19][20][21] and has been shown to improve memory function in animal models of cognitive disorders [20]. Curcumin (CU), the major bioactive component from Curcuma longa L., exerts broad and potent anti-infammatory and anti-cytokine activities [22]. Hesperidin (HES) is a favanoglycone abundantly present in citrus fruits, which exerts neuroprotective efects against PD and Huntington's disease by virtue of its antioxidant, antiinfammatory, and anti-apoptotic actions [23]. Tetrandrine (TE) is a bis-benzylisoquinoline alkaloid that is extracted from the roots of Stephania tetrandrae S. Moore. TE possesses a diverse array of biological actions, including antineuroinfammatory and antioxidative activities [24,25]. Glycyrrhizin (GLY), a triterpenoid saponin compound, is the main bioactive constituent of Glycyrrhiza glabra L., and has been shown to possess antineuroinfammatory and neuroprotective properties [26].
Combination therapy has become an emerging therapeutic strategy for complex diseases, such as neurodegenerative diseases, ofering improved clinical outcomes and reduced toxicity through a multitarget approach [26][27][28][29][30]. Te fundamental treatment principle of combination therapy is synergy, where the combined efects are greater than the sum of the individual efects [31]. Numerous studies have demonstrated synergistic or positive relationships among the bioactive compounds in complex drug combinations [31][32][33][34][35]. For instance, Park et al. [25] demonstrated that a combination of LU and L-theanine exhibited a greater efect in protecting hippocampus tissues than each compound used alone in an early-stage Alzheimer's disease animal model. Teir results showed that LU-L-theanine attenuated memory impairment and prevented tau protein phosphorylation and norepinephrine depletion in rats infused with amyloid-β in the hippocampus [36]. Although there is generally a lack of rigorous mathematical tools to accurately interpret the interaction among bioactive compounds in herbal mixtures, various methodologies have been developed to help illustrate the individual action of active ingredients in an herbal formula. Network pharmacology is a computational and mathematical model integrating literature, experimental data, and the computational sciences [37]. Recently, it has been developed to demonstrate the multitargeted actions of components in combination therapy [38]. Tis method can identify the key bioactive compounds from a complex formulation, elucidate the gene targets associated with the disease, and build the network of the bioactive compounds, gene targets, and associated signaling pathways [39]. Its increasing popularity is attributed to low research costs, a short research cycle, and comprehensive information. In drug discovery investigations, there is an emerging trend of integrating analysed information from network pharmacology with experimental results to form novel combination therapy, with a known mechanism of action [40]. Network pharmacology is usually followed by molecular docking, which can determine the binding afnity between the key bioactive and associated protein target and partly validate the predicted mode of action of the bioactive compound from the network pharmacology analysis [40]. In addition, mathematical modelling, such as combination index (CI), is widely used to quantify the interaction of drug combinations (i.e., synergistic or antagonistic) on a specifc biological target based on the experimental results [38,41].
Although the individual anti-neuroinfammatory activity of the eight selected phytochemicals has been studied, there is generally a lack of a standard approach to form synergistic combination therapy, which leads to enhanced therapeutic outcomes. Based on the versatile pharmacological actions of these eight phytochemicals associated with neuroinfammation, it is plausible that these phytochemicals may exert synergistic interactions when used in combination, leading to improved pharmacological outcomes. Tis study aimed to establish an integrative approach for developing synergistic combinations of phytochemicals using in silico network pharmacology, molecular docking, and bioassay-based validations. Our research may provide a scientifc basis and establish a new framework for novel interventions against neuroinfammation and neurodegenerative diseases.

Protein-Protein Interaction Network Construction.
Te protein-protein interaction (PPI) network describes physical interactions among protein targets that are associated with the phytochemical candidates and neuroinfammation [43]. Te STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) database (https://string-db.org/) provides the relevant protein-protein associations, which analyses confdence scores for each of the protein connections with quantifed reliability. Overlapping gene targets of each phytochemical were input to STRING to generate the PPI network [44]. Te disconnected edges were hidden in the default setting in the network, and the required interaction score was set at 0.9 as a minimum to obtain the network of PPI.

Gene Ontology and Kyoto Encyclopaedia of Genes and
Genomes Pathway Analysis. Based on the constructed PPI network of each phytochemical, the associated gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways that represent gene product properties and signaling pathways were investigated. Te Database for Annotation, Visualization, and Integrated Discovery (DA-VID) is the online enrichment analysis database (https:// david.ncifcrf.gov/home.jsp) applied to explore the relevant gene function annotation and pathway enrichment. Te GO targets were searched through the DAVID database to elucidate the interaction between related gene targets of each phytochemical and their associated GO when targeting neuroinfammation [45]. Following that, the cross-GO targets were uploaded to DAVID to obtain associated biological pathways (BP), cellular content (CC), molecular function (MF), and KEGG pathways. In the present study, the top 20 targets in each function were selected and input to Bioinformatics (https://www.bioinformatics.com.cn/) to conduct the enrichment analysis of KEGG pathways and GO enrichment used across genes, and the terms with a p value less than 0.05 were fltered for the subsequent network construction [46].

Te Construction of Compound-Gene Targets-Signaling
Pathway Network. Te compound-gene targets-signaling pathway network was constructed using Cytoscape (v.3.8.2, Institute for Systems Biology, US) [34]. Te network was built to examine the relationships among the phytochemicals, their gene targets, and associated pathways in targeting neuroinfammation. Te image of the network and associated statistics were exported from Cytoscape. Te relevant parameters were obtained by the "analysis network" tool in Cytoscape, including "Degree," "Betweenness-Centrality," "Closeness Centrality," and "Stress." Te most relevant gene targets and KEGG pathways were further fltered by selecting those "Degree" larger than 2× median values [47]. Te importance of the gene/protein target will be ranked by the degree, which will be used for the following experimental verifcation.

Molecular Docking Simulation.
In order to evaluate the credibility of the connection between the core protein targets with each phytochemical against neuroinfammation, CB-Dock (v.1.0, Yang Cao Lab, China) was used to perform molecular docking (https://clab.labshare.cn/cb-dock/php/ blinddock.php). CB-Dock is a protein-ligand docking model designed to identify binding sites, analyse center and size, and conduct molecular docking [48]. It facilitates docking procedures and increases the accuracy of molecular docking. Cavity-focused docking increases the accuracy and hits ratio with blind docking [48]. Based on the network pharmacology analysis, the structures of the selected gene targets-coded proteins were obtained from Protein Data Bank (PDB). Te crystal structures of each compound candidate were sourced from the PubChem database. Te "spacefll" and "cartoon" parameters were set for the ligand and receptor, respectively. "Element" and "chain" were used for ligands and coloured receptors.

Preparation of Selected Phytochemicals and
Lipopolysaccharides-Induced Neuroinfammation. Pure isolated phytochemicals LU, BA, AN, 6-SG, CU, HES, TE, and GLY (purity >98%), were purchased from Chengdu Bio-Purify (China). Te identity and purity were confrmed by high-performance liquid chromatography (Supplementary Material 1). Each phytochemical was dissolved in dimethyl sulfoxide (DMSO) at a concentration of 100 mM. Tey were diluted with DMEM serum-free media before adding to the  [52,53]. In the CI-Fa curve, Fa refers to the default efect level of the combination set between 0 and 1. In our study, Fa presented suppressive responses on NO from 0% to 100%. Te CI values were used to demonstrate the interaction, with CI < 1 representing synergistic interaction, CI � 1 representing no interaction (additive efect), and CI > 1 representing antagonistic interaction [30]. Te isobologram graphics were used to show synergy at three set concentrations (Fa � 0.5, Fa � 0.75, and Fa � 0.9).

Statistical Analysis.
Statistical analysis was conducted using GraphPad Prism 9.0 software (GraphPad Software Inc., USA). Te data were shown as mean ± standard error of the mean (SEM) from at least three individual experiments. Te relative IC 50 values were determined from the constructed dose-response curves. Te statistical comparison between groups was conducted by one-way ANOVA with the Tukey test, and p < 0.05 was considered statistically signifcant. where the crosstalk is likely to occur. For instance, the top common BPs of AN and 6-SG included peptidy-serine phosphorylation, sequence-specifc DNA binding transcription factor activity, LPS-mediated signaling pathway, response to stress, and positive regulation of cyclase activity. Te overlapping BPs for BA and 6-SG included positive regulation of cell growth and a LPS-mediated signaling pathway. Te common CC was late endosome and caveola for AN and 6-SG and BA and 6-SG, respectively. Tere were several common MF involved in AN and 6-SG, including MAP kinase activity, ATP binding, protein serine/threonine kinase activity, protein binding, protein kinase activity, and protein phosphatase binding. However, BA and 6-SG only showed enzyme binding as the common MF. Te top 20 KEGG pathways of each phytochemical are shown in Supplementary Material 5. Te KEGG pathways of LU, AN, 6-SG, HES, and GLY showed distinct patterns of distribution. TE only showed 11 KEGG pathways, and the pathways were at comparable degrees with no apparent distinction. CU and BA showed 9 and 3 KEGG pathways, respectively. Te top common pathway among all eight phytochemicals was identifed as the MAPK signaling pathway. In particular, the common KEGG pathway of AN and 6-SG was the MAPK signaling pathway, whereas the top KEGG pathway for BA and 6-SG was the VEGF signaling pathway. It was noticed that the MAPK signaling pathway had been shown to play a critical role in neuroinfammation [55].

Construction of Compound-Gene Targets-Signaling
Pathway Network. Te network of the eight phytochemicals was constructed with 34 nodes and 99 edges (Figure 2). Te statistical analysis from the network showed that GLY was connected with the most genes and signaling pathways (17 nodes and 16 edges). Te top fve gene targets assessed by the number of connections and degrees for the eightcompound network included MAPK14, MAPK8, NOS3, EGFR, and SRC. Te top common pathway was the MAPK signaling pathway. It was noticed that the network clearly displayed a multitarget pattern of each phytochemical that has been associated with multiple genes and signaling pathways, and they also have crosstalk as refected by the overlapping gene targets and signaling pathways.

Molecular Docking Analysis.
Based on the network pharmacology analysis, MAPK14 and NOS3 were the top two hub gene targets of all the eight phytochemicals. In order to verify this analysis, molecular docking was performed to evaluate the binding afnity between each phytochemical to MAPK14 and NOS3-coded proteins, respectively. Te binding indices of each phytochemical on MAPK and NOS3-coded proteins are shown in Figures 3(a) and 3(b), respectively. Te cavity size and afnity were evaluated using CB-Dock. Te center represents the docking pocket center coordinates. Te size parameters x, y, and z represent the directions of the docking pocket.
Te relevant indices including afnity, cavity size, and binding location (center and size) for MAPK14 and NOS3 are shown in Tables 2 and 3, respectively. Te binding energy < −5 kcal/mol was considered as the high afnity between the phytochemical and the target protein [56]. Herein, the binding afnities of the eight phytochemicals with MAPK14 and NOS3 were all less then −5, suggesting high afnities. Particularly, GLY and HES had stronger binding activities with MAPK14 (afnity <−10) than other phytochemicals, and TE showed the strongest binding with iNOS (afnity <−10).

NO Inhibitory Activity of Single Phytochemicals on LPS-Induced N11 Cells.
Te Alamar blue assay was conducted to examine the cytotoxicity of the eight phytochemicals on N11 cells. TE exhibited dose-dependent cytotoxicity from 10 to 100 μM. HES, LU, and CU showed moderate cytotoxicity from 50 to 100 μM. In contrast, AN, BA, 6-SG, and GLY did not induce any cytotoxicity from 0 to 100 μM. Te cellular neuroinfammation model was established by activating the microglial N11 cells with LPS, which led to an excessive amount of NO to 24.48 ± 0.21 ng/mL (p < 0.0001vs. blank control: 0.41 ± 0.12 ng/mL). BA, AN, LU, and 6-SG lowered LPS-induced NO expression levels in a dosedependent manner (Figure 4). Based on the dose-response curves, the IC 50 values for each active phytochemical were calculated. 6-SG showed the highest potency (IC 50   Evidence-Based Complementary and Alternative Medicine

Synergistic Efects of AN-SG and BA-SG Combinations on LPS-Induced NO Inhibition.
Te paired combinations of the eight phytochemicals were tested on LPS-induced N11 cells. Our results showed that AN-SG and BA-SG (0.50-31.34 μg/mL) appeared to be more potent than their individuals, and the enhanced NO inhibitory activities were not associated with cytotoxicity. Te rest of the pair-wised combinations did not show higher potencies than their individual components in general (Supplementary Material 3). AN-SG combination ( Figure 5(a)) exhibited a dosedependent NO inhibition with the IC 50 value of 2.85 ± 0.66 μg/mL, and it was signifcantly lower than that of AN (IC 50 = 5.22 ± 0.91 μg/mL) or 6-SG alone (IC 50 = 4.16 ± 0.46 μg/mL, both p < 0.0001). Te CI model was then performed to evaluate the drug interaction responsible for the enhanced activity in the combination. Te CI-Fa curve ( Figure 5(b)) displayed a strong synergy of AN-SG combination in inhibiting NO, with CI values ranging from 0.39 to 0.99 when the Fa was above 0.20 (20%-97% NO inhibitory efect). Te isobologram ( Figure 5(c)) also supported the observed synergy of AN-SG in reducing LPS-stimulated NO when Fa values were at 0.50 (representing 50% of the NO inhibition).
Both BA and 6-SG showed a dose-dependent inhibition of NO, as shown in Figure 5(d). Te BA-SG combination (IC 50 � 3.28 ± 0.81 μg/mL) exhibited a more prominent efect than that of BA (IC 50 � 5.48 ± 0.83 μg/ mL) or 6-SG (IC 50 � 4.16 ± 0.46 μg/mL) alone (both p < 0.0001). Te CI model revealed a synergistic interaction of BA and 6-SG when used together to suppress NO ( Figure 5(e)). At the concentration range of 3.89-8.56 μg/mL, a synergistic efect was observed with CI values ranging from 0.09 to 0.99 when the Fa was over 0.45 (45%-97% NO inhibitory efect). When Fa values were at 0.5 (representing 50% of the NO inhibition), the isobologram in Figure 5(f ) further corroborated the reported synergy of BA-SG in decreasing LPS-stimulated NO.  Figure 2: Te compound-gene targets-signaling pathway network of the eight phytochemicals related to neuroinfammation. Orange nodes represent each phytochemical candidate, blue nodes refer to potential phytochemical's targets in neuroinfammation, and the green nodes display the signaling pathway. Te size of each node represents its degree in the network. Te grey connecting lines refect that each node is interconnected.  Evidence-Based Complementary and Alternative Medicine

Compound-Gene Targets-Signaling Pathway Networks of AN-SG and BA-SG Combinations in Relation to
Neuroinfammation. Te compound-gene targets-signaling pathway networks of the AN-SG and BA-SG combinations were built to understand the associated mechanisms of their synergistic interaction. As shown in Figure 6, MAPK14 and NOS3 appeared to be the top hub gene targets, and the MAPK signalling pathway is the top overlapping pathway for both paired combinations. Ten, the modulatory efects of these two combinations in comparison with their individual component on MAPK14 and iNOS protein expressions were examined by Western blot analysis.

Synergistic Mechanisms of AN-SG and BA-SG Combinations in Relation to Modulated Phosphor-MAPKp38/ MAPKp38 and iNOS Protein Expression Based on the Network Pharmacology Analysis.
To verify the network pharmacology analysis and investigate the synergistic mechanism of AN-SG and BA-SG combinations, the protein levels of phosphor-MAPKp38/MAPKp38 (p-p38/p38) and iNOS were investigated. As shown in Figures 7(a) and 7(b), the stimulation of LPS (1 μg/mL) led to signifcantly upregulated expressions of p-p38/p38 (p < 0.0001) and iNOS (p < 0.001) with fold increases of 2.06 ± 0.22 and 4.16 ± 0.08 in comparison to that of the untreated cells (Blank), respectively. AN, 6-SG, and AN-SG all signifcantly inhibited the increased fold change of p-p38/p38 (p < 0.05vs. LPS) and iNOS (p < 0.05vs. LPS). In addition, the inhibitory efect of the AN-SG combination was signifcantly greater than that of AN or 6-SG alone on iNOS (p < 0.05vs. AN or 6-SG).

Discussion
Network pharmacology analysis has emerged as a powerful tool for the development of combination therapy [57] and is particularly useful in Chinese herbal medicine research to understand the multitargeted mechanisms of the bioactive components in complex herbal formulations [58]. Herein, network pharmacology has been applied to evaluate the pharmacological actions of eight phytochemicals selected from Chinese herbs targeting neuroinfammation. Molecular docking and experimental bioassays were followed to explore synergistic interactions and the associated mechanisms of the pair-wised combinations among eight phytochemicals. Our results demonstrated that two paired combinations exhibited synergy in inhibiting LPS-induced NO production on microglia N11 cells, and the mechanisms were associated with the downregulation of MAPK p-p38/ p38 and iNOS protein expressions. Te data obtained from the experimental study is in line with the illustration from the network pharmacology and molecular docking analysis.
Te highly relevant gene targets for the antineuroinfammatory efects of the eight phytochemicals included MAPK14 and MAPK8. Te top KEGG pathway involved in the actions of the eight phytochemicals was the MAPK signaling pathway. MAPKs are a family of serine/threonine protein kinases that regulate key biological processes as well as cellular responses to external stress signals [59]. MAPKs are vital for intracellular signal transduction and play critical roles in regulating cell proliferation, brain plasticity, infammatory responses, and other biological functions [60]. According to recent preclinical studies, increased MAPK activation is a signifcant factor in brain infammation. p38α/MAPK14 and extracellular signal-regulated kinase (ERK) are intracellular signaling regulators [61,62], which mediate the expression of the iNOS and TNF genes in LPSactivated glial cells, suggesting the role of p38MAPK in the activated glial cells [63,64]. Increasing evidence showed that MAPK cascades, including mitogen-and stress-activated kinase 1 and mitogen-and stress-activated kinase 2, were associated with the production of IL-1 in the BV-2 mouse microglia cell line and primary rat microglia [65]. MAPK14 is ubiquitously expressed and plays an important role in proinfammatory signaling, making it an appealing therapeutic target for chronic infammatory diseases [66]. Antineuroinfammatory therapies might be directed by targeting MAPKs kinases, such as MAPK p38 and their role in the transcription and translation of infammation mediators, and can lead to an enhanced therapeutic outcome [67]. Our network pharmacology fndings demonstrated that the  Evidence-Based Complementary and Alternative Medicine MAPK signaling pathway was the common relevant pathway linking the actions of the phytochemicals to neuroinfammation, which correlates with previous fndings. Nitric oxide synthases (NOS) are the enzymes responsible for NO generation. To date, three distinct NOS isoforms have been identifed, including neuronal NOS (nNOS/NOS1), inducible NOS (iNOS/NOS2), and endothelial NOS (eNOS/NOS3) [68]. Te excessive NO is one of the important neuroinfammatory mediators that trigger neuronal toxicity and death [69]. A study by Connelly et al. [72] suggested that mice with macrophages NOS3 knock-out revealed downregulated nuclear factor kappa B (NF-κB) signaling and reduced expression of iNOS, resulting in decreased LPS-induced NO generation and ultimately suppressed neuroinfammatory response [70]. Tus, these studies supported the results from our network pharmacology analysis that NOS3 is an important mediator in neuroinfammation. Moreover, iNOS was generated in activated microglial cells and mediated NO synthesis [71]. Studies in patients with Parkinson's disease revealed an increased density of iNOS-expressing glial cells in the substantia nigra compared to the control [72]. In addition, iNOS has been linked with microglial activation, inducing an infammatory response and resulting in neural cell death [73]. Apart from the most relevant disease proteins and therapeutic targets discussed above, other neuroinfammation targets including EGFR, SRC, CASP3, and PPARG in the compound-gene targets-signaling pathway networks may potentially participate in the  pharmacological actions of the eight phytochemicals against neuroinfammation. Molecular docking is commonly used to model the interaction between small molecules and target proteins [74] and to estimate the binding energy of a ligand and the intensity of the interactions. Tus, molecular docking has the capacity to identify novel phytochemicals of therapeutic interest, predict the ligand-target interactions at a protein level, and determine the degree of binding afnity between a phytochemical and its target proteins [75]. Our results suggested that all eight phytochemicals exhibited strong binding afnity with MAPK14 and NOS3 according to intermolecular interactions. Tis result indicated that these phytochemicals displayed good binding activities with the hub targets of neuroinfammation, illustrating the mechanisms of action underlying the therapeutic efects of these phytochemicals [76]. Our molecular docking results provide evidence to support the network pharmacology fndings at the molecular level.
In vitro bioassays were carried out to validate the results of the network pharmacology analysis and to identify possible synergistic interactions among the eight phytochemicals. LPS-induced microglia N11 cells were used to test the individual and combined activities of the eight phytochemicals in inhibiting NO production. Our results suggested that LU, AN, BA, and 6-SG lowered LPS-induced NO expression levels in a dose-dependent manner. Interestingly, although HES and GLY showed high binding afnities to both MAPK14 and NOS3 proteins, they did not show any obvious inhibitory efect in the NO assay, which may be attributed to their broad and unspecifc efects. Furthermore, our study demonstrated that the AN-SG and BA-SG combinations exhibited synergistic interaction in reducing NO production. To the best of our knowledge, it is the frst study demonstrating synergistic interactions between AN and 6-SG and BA and 6-SG targeting neuroinfammation. Based on the constructed network, the MAPK14 and NOS3 gene targets were highly relevant for the anti-neuroinfammatory actions of the AN-SG and BA-SG combinations, both linking to the downstream production of iNOS protein expression [77,78]. Our Western blot analysis suggested that the synergistic mechanisms may be associated with downregulated phosphor-MAPKp38 and iNOS proteins. Tese fndings are in line with the predictions from our network pharmacology analysis that the crosstalk among AN, 6-SG, and BA were likely to occur on their hub gene targets, MAPK8, MAPK14 and NOS3, and their coded proteins.
Previous studies suggested that AN, BA, and 6-SG exhibited anti-neuroinfammatory activity via the downregulation of NF-κB and MAPK signalling pathways. AN downregulated the protein levels of iNOS and cyclooxygenase-2 (COX-2) and mRNA expression in LPSinduced human keratinocyte HaCaT cells [79]. BA signifcantly inhibited the activation of MAPKs and suppressed the transcriptional activity of NF-κB [80]. Similarly, treatment with 6-SG resulted in the reduction of LPS-induced NF-κB subunit and the dependent transcriptional activity of NF-κB by blocking the phosphorylation of NF-kappa-B inhibitor alpha(IκBα) and subsequent degradation of IκBα [81]. 6-SG also interferes with the activation of PI3K/Akt/IκB kinases (IKK) and MAPK [20]. Tus, the common signaling pathways of NF-κB and MAPK may be the key to the synergistic mechanism of the AN-SG and BA-SG combinations.
Te current study only focused on analysing the MAPK pathway based on the network pharmacology results. More in-depth studies of other hub KEGG pathways (i.e. NF-κB) and their target proteins by the two combinations will be conducted in the future. Moreover, it is evident that AN, 6-SG, and BA were associated with many other key gene targets and coded proteins, as displayed by the compound-gene target network. Tese gene targets and signalling pathways may also contribute to the synergistic efects of AN-SG and BA-SG against neuroinfammation.
Te network pharmacology replies primarily to the currently available data from online databases, which may lead to a biased result [82]. Terefore, it should only be used as a predictive tool for the molecular mechanisms of a compound mixture and reveal the relevant relationship between the phytochemical and gene targets without providing further information on their downregulatory or upregulatory actions. To elucidate the detailed mode of action, it is no doubt that more experimental investigations should be carried out. We also recognise that the cellular assays used in this study serve as a primary screening method. Whether the AN-SG and BA-SG can be applied to a broader range of neuroinfammatory mediators and reserve the synergistic actions in a whole organism and a clinical setting are yet to be determined. Also, it is unclear if synergy occurs in the high-order combinations among eight phytochemicals which requires further investigation.

Conclusions
Te present study developed two novel herbal compound combinations, AN-SG and BA-SG, that showed synergistic anti-neuroinfammatory activities. Te associated mechanisms underlying the observed synergy were explored through network pharmacology analysis and molecular docking. Network pharmacology demonstrated that MAPK14, MAPK8, and NOS3 were the main gene targets of AN, 6-SG, and BA, and the top KEGG pathway was the MAPK signaling pathway. Western blot analysis demonstrated that AN-SG and BA-SG showed prominent efects in inhibiting p-p38 MAPK and iNOS, which has partly validated the illustration from network pharmacology. Te present study provides insight into the development of synergistic combinations targeting neuroinfammation with integrated network pharmacology, molecular docking, and experimental bioassays.

Data Availability
Te analysed data used to support the fndings of this study are included within the article, and the raw data used to support the fndings of this study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that there are no conficts of interest.