A network pharmacology-based analysis to distinguish the differences between Lonicerae Japonicae Flos and Lonicerae Flos

Multiple basal plants are commonly used as materia medica in the traditional medicine of various nationalities and ethnicities worldwide. We call this practice “multibasal-plant materia medica” (MBPMM). So we proposed the application of network pharmacological method that it can provide a new way of distinguishing the differences among the different basal plants used in traditional medicines. We apply the method in investigating the differences and similarities in the material bases and mechanisms of anti-inammatory activities of Lonicerae Japonicae Flos and Lonicerae Flos. Lonicerae Japonicae Flos and Lonicerae Flos share plenty of similarities in terms of anti-inammatory mechanisms and material bases. Both of them mainly act on airway inammation and tumour inammation via the NF-κB signalling pathway and immune response, oxidation and signal transduction. However, Lonicerae Flos acts on inammation with greater intensity than Lonicerae Japonicae Flos. We argue that they can be used interchangeably for the prevention and treatment of tumours and airway inammation at a proper dosage. Otherwise, Lonicerae Flos may be more appropriate for treating neurological and metabolism-related inammation, whereas Lonicerae Japonicae Flos is more suitable for the treatment of inammation of systemic organs, such as intestines.


Introduction
Traditional medicine is a valuable experience that people have gained in their struggle against nature and diseases. People derive bene ts from traditional medicines, such as artemisnin and reserpine [1,2]. As one of the three systems of traditional medicine, traditional Chinese medicine (TCM) has been an important part of health care systems in many countries, especially in Asia. TCM has good clinical treatment and disease-prevention effects; thus, it has gained widespread recognition worldwide [3].
The use of "multibasal-plant materia medica" (MBPMM) is a widely accepted practice in TCM. MBPMM is recognised in various pharmacy books and health literature, such as the Pharmacopoeia of the People's Republic of China [4]. However, different species have different chemical compositions that exert different pharmacodynamics. For example, Asari Eadix Et Rhizoma is a typical TCM that includes three basal plants, two of which are Asarum heterotropoides Fr. Schmidt var. mandshuricum (Maxim.) Kitag. and A. sieboldii Miq. Experiments demonstrated that A. heterotropoides var. mandshuricum has a stronger anti-in ammatory activity than A. sieboldii Miq [5]. The base plant of LJF is Lonicera japonica Thunb., which is mainly found in northern Chinese provinces, such as Shandong, Shaanxi and Henan. The main source plant of LF is L. macranthoides Hand.-Mazz. In addition, LF also include L. hypoglauca Miq., L. confuse DC. and L. fulvotomentosa Hsu et S.C. Cheng. These plants are mainly found in Sichuan, Hunan and Guangdong Provinces in southern China [6,7].
These basal plants belong to the same genus. The two herbs are identical in nature and avour xiong, meridian tropism, directions and dosage. They can treat sores, rooted sores, wind-heat common cold and weakness (Chinese Pharmacopoeia Commission, 2015). In modern pharmacology, both LJF and LF have anti-in ammatory, antibacterial, antiviral, antioxidant and hepatoprotective pharmacological effects [8][9][10][11]. However, several studies have indicated that LJF and LF have different pharmacological effects. LJF has a wider range of antimicrobial activity than LF. The LD 50 values of the two herbs are not considerably different, but the potential toxicity of LF may be greater than that of LFJ because of the former's hypersensitivity reaction [12]. At present, comparative studies on LJF and LF are reductionist and mostly conducted in terms of the content of their characteristic components or speci c pharmacological activities.
The network pharmacology approach is based on systems biology and multidirectional pharmacology.
This approach constructs the relationship between active compounds and the organism from the perspective of component-target-pathway-disease. This approach is mostly used to explain the characteristics and mechanisms of multicomponent, multitarget and multipathway actions in TCM [13].
This approach has been successfully adopted in TCM research [14]. In this study, the network pharmacology method was employed to investigate the differences between LJF and LF. We adopted the network pharmacology method to compare the differences and similarities in the pharmacological effects of LJF and LF. The results provided a reference for the combined use of these herbs. Speci cally, we examined the anti-in ammatory pharmacological activities of LJF and LF and evaluated whether they can be used as substitutes for each other. This study provides new perspectives for research and addressing similar problems in TCM.

Establishment of Chemical Composition Database
We retrieved information on the chemical compositions of LJF and LF from the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform and The Encyclopaedia of Traditional Chinese Medicine [15,16]. These databases are a repository of chemical information of traditional Chinese herbs. Moreover, we collected additional information about these herbs from the literature [17,18]. Then the free chemical information online website Chemspider [19] and PubChem has been used to translate and check the le type database [20]. Finally, MOE 2019 software was been choosed to merge the data and establish a database.

Target Fishing
Most TCM compounds exert corresponding biological functions by acting on protein targets, which simultaneously induce a series of physiological change. Identifying the targets is meaningful to understand the mechanism of a compounds' action. In this study, we employed the SEAware and Swiss Target Prediction webserver for the target shing of the compounds we collected [21]. The target shing methods we performed were based on 'similarity hypothesis' of similarity among small molecules. This hypothesis states that compounds with similar structures have similar physical and chemical properties and biological activities. The activity and targets of unknown molecules can be predicted by comparing the small molecules with activity data [22]. We calibrated the names of the targets to the o cial gene name through the Uniprot database [23].

Construction of Disease Target Database
We consulted several databases for information on in ammation-related targets. Genecard database assembles data from 150 websites and provides comprehensive genomic, proteomics, genetic, clinical and functional information [24]. We used the keywords 'in ammation' and 'in ammatory' in the retrieval, and we collected the genes with high scores (≥ 10) to build a database of targets related to in ammatory diseases. Finally, we integrated the targets from target shing with the disease target database, and then the overlapping parts as the in ammation-related targets of LJF and LF were selected.
STRING [25] was used to obtain PPI network data of LJF and LF with high credibility score (≥ 0.7), and then these data were putted into Cytoscape [26]. We used Degree to indicate the number of connections between a node and other nodes and Betweenness Centrality to re ect the value of the bridge centrality of the node. These parameters are crucial indicators to screen, obtain and compare the core target relationship network of two Chinese medicines [27].

Enrichment Analysis
We used the Database for Annotation, Visualisation and Integrated Discovery [28] to analyse the Gene Ontology (GO) and Kyoto Encyclopaedia of Genes and Genomes (KEGG) of in ammation-related genes.
GO enrichment analysis included biological process (BP), molecular function (MF) and cellular component (CC). We determined and compared the anti-in ammatory mechanisms of LJF and LF.

Comparison of Key Compounds
Using degree as reference values, we compared the key compounds that act on the anti-in ammatory targets of LJF and LF on the basis of their types and related features.

Collection of Compounds and Targets
The databases of LJF and LF contained 243 and 200 compounds, respectively, 66 of which were common to both herbs (Supplementary Table 1). A total of 207 targets for LJF and 198 targets for LF were obtained as potential targets after screening and weight reduction, of which 176 were the same. A total of 1,006 genes related to in ammation were collected in GeneCards. We integrated the compound and in ammation targets to obtain the common targets with in ammation. Fifty-seven targets were identi ed from both LJF and LF, 49 of which were shared by them (Fig. 3).

Analysis of Key Target Network
We introduced the in ammation targets of the two herbs to STRING. To illustrate the strength of the correlation between proteins, we acquired two PPI maps after setting the con dence level higher than 0.7. After ltering using the cut-off value, the PPI of LJF consisted of 44 nodes with 90 edges, whereas that of LF contained 48 nodes with 105 edges. The key targets were screened by analysing the values of Degree and Betweeness Centrality of each target in the respective networks and by using their medians as thresholds (Fig. 4). Results showed that LJF and LF had nine core targets. Six common targets, namely, APP, VEGFA, MMP9, EGFR, PPARG and ALOX5, were found. Information on the 12 target genes is listed in Table 1.  Fig. 5.
LJF and LF mainly act in extracellular space, including lysosomes and other cell components, for in ammation. The targets of LJF were located on the external side of plasma membrane, including pivotal proteins with anti-in ammatory properties, such as ApoA-I [29]. The I-κB/NF-κB complex was found to be more relevant to LF than to LJF.
In MF, the two drugs speci cally have RNA polymerase II transcription factor activity and combine various molecular functions, including enzyme, chromatin binding, transcription factor and other molecular functions. Furthermore, LJF has special functions related to chemokine receptor activity, whereas LF has peroxidase activity. We analysed the involved biological processes and divided the co-participatory processes into ve categories: apoptosis and proliferation (GO: 0022617, GO: 0043066), direct in ammation (GO: 0006954, GO: 0019369), immunity (GO: 0019372), oxidation (GO: 0055114) and signal transduction (GO: 0070374). These processes are directly or indirectly related to the development of in ammation. A total of 11 and 18 pathways of LJF and LF, respectively, were screened by KEGG pathway, of which nine terms were the same for both herbs. Two 'target-path' combination networks of LJF and LF were constructed (Fig. 6). The circle represents the corresponding anti-in ammatory target, whereas the triangle denotes the enrichment pathway. Nodes depicted in blue indicate the unique targets/pathways of LJF and LF. LF showed more characteristic pathways than LJF. The targets of LJF and LF in the same pathway were not exactly the same. The pathways common to both herbs included arachidonic acid metabolism, in ammatory mediator regulation of TRP channels and tumour-related pathways, which are closely related to in ammation. In addition, LF has a unique TNF signalling pathway, and HIF-1 signalling pathway is related to in ammation.

Comparison of Compounds
We analysed the compounds of LJF and LF and their relevant anti-in ammatory targets. After that, eight and seven key compounds from LJF and LF has been obtained, respectively. These compounds could be divided into avonoids or organic acid. Three groups of LJF and LF compounds, namely, J185 and S144, J179 and S141 and J180 and S142, have the same structures (Fig. 7).
Due to technical limitations, Table 2 is provided in the Supplementary Files section.
The compounds corresponding to anti-in ammatory targets were classi ed according to the main types of these herbs. As shown in Fig. 8, the class of each compound (organic acids, triterpenes and others) was compared, showing that the structures of these compounds only slightly differ. The quantity and types of avonoids in LJF were more plentiful than those in LF.

Discussion
Researchers debate the differences in pharmacological effects of LJF and LF. Both claims are supported by experimental data. Existing comparative studies are incomplete and not systematically designed. In the present study, we adopted a network pharmacology approach to compare the differences in pharmacological mechanisms of two basal plants with anti-in ammatory activity. We hope to provide a reference for the comparative study of LFJ and LF.
We employed a common screening criterion to compare the protein-protein interaction network of LJF and LF. Among the nine core genes obtained, six were common in both herbs, accounting for 66.7%. We investigated each gene separately (Table 3). Results showed that both LJF and LF mainly act on airway in ammation and tumour-induced in ammation. Both herbs were mainly involved in the NF-κB signalling pathway for anti-in ammatory effects. The extract of LF blocks the activation of the NF-κB in ammatory signalling pathway by inhibiting IκBα phosphorylation with NF-κB p65 and IκBα degradation [30]. The extract of LJF substantially reduces p50 and IKK expression levels on the NF-κB pathway [31]. Therefore, LJF and LF act synergistically on the NF-κB signalling pathway but under different mechanisms. LJF has a considerable effect on ovalbumin-induced asthma in a rat model [32]. By contrast, research on LF is few and there is no study has directly demonstrated that whether the LF have an proportional effect on asthma or not. According to the results of the analysis of speci c genes, LF is more correlated with the NF-κB pathway and has an effect on neuro-in ammation, whereas LJF is more biased towards systemic in ammation, such as enteritis, pneumonia and in ammation by microbial infection. Table 3 Information on core targets in relation to anti-in ammatory genes.

Attributes Of The Gene Target
Gene Name In ammatory-Related Mechanisms/Etiology/Disease Common APP Activates individual nuclear phagocytes in the brain and causes an in ammatory response Activated TLR4 signaling; Common VEGFA Airway in ammation such as asthma [33]; Tumor-induced in ammation; Common MMP9 Airway in ammation [34] Suppression of malignancy through in ammation [35]; Common EGFR Possible activation of NF-kappa-B signal; Airway in ammation [36];

LJF MPO
In ammation caused bymicroorganisms [44] in ammation of the lungs [45]; GO and KEGG enrichment analyses of the gene targets revealed that both LJF and LF exert antiin ammatory biological activity in a multitargeted multipathway manner. The targets of both herbs are mostly distributed in the extracellular gap and involve ve types of biological processes and associated in ammatory pathways. Various substances, such as cyclophilin A and exosomes, are associated with in ammation in the extracellular gap. Exosomes play an important regulatory role in the development of chronic in ammatory airway disease [46]. Among the ve major biological processes, apoptosis and aberrant proliferation are the most important processes and features of tumorigenesis, such as extracellular matrix decomposition (GO: 0022617), which breaks down extracellular matrix and multiple cytokines that affect cell survival and regulate cell proliferation, thereby avoiding acute or chronic in ammatory damage caused by tumours. Thus, LJF and LF are closely associated with tumours and in ammation, a conclusion that is consistent with the results of core target analysis. Two main aspects are involved in the immune process, namely, the in ammation caused by immune abnormalities and the immune response that accompanies the in ammatory response. The main pathological response of the former is autoimmune disease and tumour-associated in ammatory response. LJF and LF can exert antiin ammatory effects by regulating the root cause of in ammation, that is, immunity. In the latter, LJF and LF assist in decreasing the in ammatory response by activating or enhancing the immune response by releasing anti-in ammatory factors. Both oxidation and signal transduction are also directly related to the occurrence and regulation of in ammation. These features are the similarities in anti-in ammatory mechanisms found in the GO and KEGG enrichment analyses of LJF and LF. These similarities can be visualised from the bubble and network diagrams. Their GO entries are mostly the same, although the mechanism of LJF and LF is different.
When it comes to the pathway of LJF and LF, here are some similarities. Figure 8 is one of the representatives. But more often, the GO results shows LJF and LF are not involved in exactly the same targets, such as in ammatory response (GO: 0006954) and Fig. 6. Hence, LJF and LF probably act in different ways on the same pathway and end up acting at different intensities because of the different correlations between the targets. In addition, given that LF contains more anti-in ammatory-related entries, we analysed the characteristic targets of the source of this phenomenon. LF has more characteristic targets of action in the immune pathways than LJF, and both are neurologically involved or related to fat metabolism [47,48]. We inferred that the anti-in ammatory effects of LF are more intense than those of LJF, as well as targeted more to neurological and metabolic aspects of in ammation. However, no relevant studies are available to support this conjecture, and pharmacological experiments should validate these results.
We further analysed the material basis of the anti-in ammatory activity of LJF and LF. We found many similarities and differences. The types of central compounds of both LJF and LF are avonoids and organic acids, which are the main compounds that exert anti-in ammatory effects. Flavonoids are recognised for their anti-in ammatory activity, and the organic acids in LJF and LF have been shown to be the main anti-in ammatory components [49][50][51]. Moreover, three other avonoids, namely, luteolin, quercetin and kaempferol, were identical in the central compound. These compounds are reportedly the major components of LJF (in large amounts) and have clear anti-in ammatory activity [52,53]. The results revealed the similarity in substance bases of the two herbs and demonstrated the effectiveness of the screening method. The pharmacological activities of the other central compounds of LJF and LF are not reported in the literature. Hence, future studies should investigate these aspects. Although the compounds are different, their biological activities are similar because their structures are similar. This condition leads to a large target repetitive rate. With regard to compounds that correspond to common targets associated with in ammation, LJF has more avonoids than LF. The other chemical types were not substantially different. The content of chlorogenic acid analogues in LF is substantially higher than that in LJF [54]. We hypothesise that the material differences between LJF and LF are mainly due to considerably different avonoids and organic acids.

Conclusion
In this study, the mechanism of action of two Chinese herbal medicines and the material basis of their anti-in ammatory effects were investigated via a network pharmacology approach. The targets of LJF and LF were predicted. LJF and LF were found to have many similarities in terms of mechanisms and material bases. Both them mainly act on airway in ammation and tumour in ammation and in uence the NF-κB signalling pathway. Their mechanisms involve ve types of biological processes, such as apoptosis and proliferation, direct in ammation-related, immune response, oxidation and signal transduction. Given that LF has a richer material base, it may act on in ammation with greater intensity than LJF. Therefore, we suggest that LJF and LF can be substituted for each other for the prevention and treatment of tumours and airway in ammation at an appropriate dosage. LF is likely more appropriately used in neurological and metabolism-related in ammation, whereas LJF is more suitable for the treatment of in ammation of systemic organs, such as intestines and lungs. The results provide a reference for addressing the problems with MBPMM. Network pharmacology may be adopted to study similar issues.