Identifying Active Compounds and Mechanisms of Citrus changshan-Huyou Y. B. Chang against URTIs-Associated Inflammation by Network Pharmacology in Combination with Molecular Docking

Purpose The ripe fruits of Citrus changshan-huyou, known as Quzhou Fructus Aurantii (QFA), have been commonly used for respiratory diseases. The purpose of this study was to investigate their active compounds and demonstrate their mechanism in the treatment of upper respiratory tract infections (URTIs) through network pharmacology and molecular docking. Methods The prominent compounds of QFA were acquired from TCMSP database. Their targets were retrieved from SwissTargetPrediction database, and target genes associated with URTIs were collected from DisGeNET and GeneCards databases. The target protein-protein interaction (PPI) network was constructed by using STRING database and Cytoscape. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) were enriched. Visual compound-target-pathway network was established with Cytoscape. The effects of compounds were verified on the inhibitory activities against phosphoinositide 3-kinases (PI3Ks). Finally, the molecular docking was carried out to confirm the binding affinity of the bioactive compounds and target proteins. Results Five important active compounds, naringenin (NAR), tangeretin (TAN), luteolin (LUT), hesperetin (HES), and auraptene (AUR), were obtained. The enrichment analysis demonstrated that the pathways associated with inflammation mainly contained PI3K/Akt signalling pathway, TNF signalling pathway, and so on. The most important targets covering inflammation-related proteins might be PI3Ks. In vitro assays and molecular docking exhibited that TAN, LUT, and AUR acted as PI3Kγ inhibitors. Conclusion The results revealed that QFA could treat URTIs through a multi-compound, multi-target, multi-pathway network, in which TAN, LUT, and AUR acted as PI3Kγ inhibitors, probably contributing to a crucial role in treatment of URTIs.


Introduction
Upper respiratory tract infections (URTIs) include nasopharyngitis (common cold), sinusitis, pharyngitis, laryngitis, and laryngotracheitis, which is a common infection in children generally caused by viral, respiratory infection in the mouth, nose, throat, larynx (voice box), and trachea (windpipe) [1]. Complications or deaths due to virus infections are often associated with inflammation events, which cause more harm in children. For example, the seasonal influenza A virus (IVA) infects the respiratory tract, causing epithelial damage, pulmonary infiltration, hypoxemia, and even leading to acute respiratory distress syndrome (ARDS) [2]. Currently, there still a lack of antiviral drugs specifically for the treatment of URTIs, although analgesics and antipyretics are benefit for relieving symptoms such as pain and fever. Antibiotics are ineffective to treat viral infections, and inappropriate use of them for URTIs in Chinese children remains rampant [3]. erefore, it is urgent to reduce antibacterial treatment and use symptomatic drug properly, especially for child patients [4]. Traditional Chinese medicine (TCM) was widely used for prevention and treatment of URTIs. Accumulating evidence has indicated that anti-inflammatory effect of TCM plays an important role in the treatment process, through nuclear factor kappa-B (NF-κB), phosphoinositide 3-kinase (PI3K)/Akt, extracellular signal-regulated kinase (ERK), and signal transducer and activator of transcription 3 (STAT3) signalling pathways [5][6][7].
Quzhou Fructus Aurantii (QFA), formerly known as Changshan Huyou, the ripe fruits of Citrus changshanhuyou Y. B. Chang, are commonly used for respiratory diseases (such as dry cough after catching a cold) and digestive system diseases [8,9]. According to Quzhou Prefecture Chronicles and Changshan County annals, the medicinal history of QFA can be traced back to the Qing Dynasty. With the time going, it was officially included in Zhejiang Traditional Chinese Medicine Processing Standards (2015 edition), as one of the genuine medicinal materials of new "Zhebawei" in Zhejiang Province and "Quliuwei" in Quzhou City, Zhejiang Province of China. Pharmacological research indicated that the water extract of QFA had antitussive and expectorant effects [10]. Recently it was reported that QFA extracts had anti-inflammatory effect on acute lung injury (ALI) [11] and could also prevent obesity, as well as associated metabolic diseases, such as hyperlipidemia and diabetes [12][13][14].
e major active components of QFA were considered to be flavonoids, alkaloids, and volatile oils. e flavonoids might be the main ingredients in treatment of URTIs which displayed antiinflammatory effects [15,16]. However, the specific compounds that exert these effects and their molecular mechanisms remain unclear.
With the support of the local government, the research and development of QFA are highly expected. TCM is characterized by comprehensive medical effects with complex matrices and multiple therapeutic targets.
erefore, it is difficult to elucidate the underlying molecular mechanisms. Compared with experimental research, computational biology research is an option to identify targets and signalling pathway in less time. In the present study, network pharmacology was used for evaluating molecular mechanisms by analyzing the main active compounds and targets of QFA. e inhibitory activities of main active compounds on targets were verified in vitro. At last, molecular docking was also applied to observing the binding affinity of ligand-target to confirm the inhibitory active compounds and targets.

Screening Compounds of QFA.
e PubChem database (https://pubchem.ncbi.nlm.nih.gov/) was used for retrieving the 2D chemical structure. Traditional Chinese Medicine Systems Pharmacology Database with Analysis Platform (TCMSP, https://old.tcmsp-e.com/tcmsp.php/) provided pharmacokinetic properties for natural compounds involving oral bioavailability, drug-likeness, intestinal epithelial permeability, blood-brain barrier, and aqueous solubility. By entering the molecule name contained in QFA in the search box to search, the compounds with oral bioavailability (OB) ≥20% and drug-like components ≥0.18 were displayed. en the active compounds were analyzed to identify biological target genes in the SwissTargetPrediction database (https://www.swisstargetprediction.ch/).

Protein-Protein Interaction (PPI) Network and Compound-Disease-Target Network.
e potential targets were input into STRING database (https://www.string-db.org/) to obtain the targets PPI network. Cytoscape 3.7.1 was applied to constructing the PPI network and Compound-Disease-Target network. Network Analyzer, a network topology analysis plug-in in Cytoscape, was used for topological analysis of PPI networks.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway Enrichment Analyses.
e potential targets were input into Metascape databases (https:// metascape.org/) and selected the species as "Homo sapiens." To perform GO, threshold was set to a p value <0.01, a minimum count of 3, and an enrichment factor >1.5 [17]. KEGG function was used for doing pathway analysis, and the top 10 KEGG pathways with a p value <0.01, a minimum count of 3, and an enrichment factor >1.5 were selected.

Construction of the Compound-Target-Pathways
Network of QFA. Visual compound-target-pathway network was established with Cytoscape 3.7.1 to reflect the complex relationships. Network Analyzer was used for analyzing the network. e degree value represents the number of connections between those nodes. e size of node is related to the degree value. e larger the value is, the more interrelated the compound, target, or pathway is. . After shaking the plate, 2.5 μL of substrate solution was added to each well to start reaction, and the final concentrations of PIP2 and ATP were 50 μM and 25 μM, respectively. e assay plate was covered and incubated at room temperature for 1 h. 5 μL of ADP-Glo reagent was then added to each well to stop the reaction. Subsequently, the mixture was treated briefly with centrifuge rotor (Avanti J-15R, Beckman Coulter, SN: JBR20A055), shaken slowly, and equilibrated for 120 min. en, 10 μL kinase detection solutions was add to each well, followed by equilibration for 30 min before reading on a plate reader for luminescence. Finally, conversion data was collected on Envision (2104 Multilabel Reader, Perkin Elmer, SN: 1041048) and relative light unit (RLU) values were converted to inhibition values using the formula of (sample RLU-min)/ (max-min) × 100. Herein, "min" means the RLU of no enzyme control and "max" means the RLU of DMSO control. e results were shown as graph in percentage of kinase inhibition in comparison with positive and negative control, and the IC 50 values were calculated.

Molecular Docking Analysis.
e 3D structure of active substance of QFA was modelled by utilizing the sketcher toolbars within SYBYL-X 2.0 software. e receptor was searched from the RCSB PDB database (https://www.rcsb.org/). e docking calculations were performed by the Flex Dock program within SYBYL-X 2.0. PyMOL software (the PyMOL Molecular Graphics System, Version 2.0, Schrodinger, LLC.) was used for performing ligand/receptor analysis.
Six compounds corresponded to 92, 103, 103, 9, 103, and 113 targets from the SwissTargetPrediction database, respectively. After removing the duplicated genes, a total of 287 targets were finally retrieved from the Swis-sTargetPrediction database.

Predicting Target Proteins for
URTIs. 1951 possible treatment targets were retrieved after removing duplicated targets from DisGeNET and GeneCards database. Further, by Venny 2.1 drawing software, the 287 druggable targets were mapped to 1951 URTIs-related disease targets. Finally, 117 target proteins of corresponding active compounds of QFA were identified to be potential targets in treatment of URTIs. Compound-Disease-Target network is shown in Figure 1.

PPI Network.
e 117 hub genes were further analyzed on the STRING online data platform and network establishment, and then the data was input into Cytoscape 3.7.1 to obtain the PPI network ( Figure 2), which contained 113 nodes and 627 edges. e average node degree was 11.9 after analysis, and the median node degree was 9. Each circular node represented a protein target in the network. e degree value was represented by the number of lines connected to the same node, which means the importance of each node in the network. e larger the node in the PPI network is, the greater its degree value is. Each edge represented the interaction between proteins in the PPI network. e more lines there are in the PPI, the closer associations are found in the drug and the disease.
ere were 19 target proteins whose degree value is more than twice the median value ( Figure 2, right). e degrees of AKT1, ESR1, SRC, EGFR, PTGS2, and MMP9 (58, 43, 43, 40, 38, and 33, respectively) were significantly higher than those of the other targets, suggesting that these six targets were the most important targets for the treatment of URTIs.

GO and KEGG Pathway Enrichment
Analyses. GO terms were selected according to the p value parameter (p ≤ 0.01). To investigate integral regulation of URTIs-associated inflammation by QFA, a total of 299 pathways were obtained from database. e top 10 channels are displayed in Figure 3(d). e main process associated with inflammation included PI3K/Akt, tumour necrosis factor (TNF), and NF-κB signalling pathways.

Compound-Target-Pathway
Network. e compoundtarget-pathway network was performed by Cytoscape3.7.1 software, which contained 88 nodes (6 compound nodes, 72 targets, and 10 pathways) and 325 edges in total (Figure 4). It showed that each active compound could act on multiple targets and pathways. Based on the degree of compoundtarget-pathway network, five compounds, TAN, LUT, HES, NAR, and AUR, with the degree values of 36, 33, 32, 24, and 18, respectively, had larger nodes, indicating that they were probably the main active ingredients. However, the degree value of NOB was only 2. MET, MDM2, PIK3CA, IGF1R, and MMP9 were the top 5 targets, with the degree values of

Inhibitory Activities of PI3Kα, β, c, and δ.
Experimental effect of five main active compounds on PI3K/Akt signalling pathway has been verified by assaying their inhibition of PI3Ks in vitro. e IC 50 values of five compounds towards Class I α, β, δ, and c isoforms of human PI3 kinases were determined in the range of 0.625 to 40 μM by using Promega ADP-Glo Kinase Assay Kit. As shown in Table 2 Figure 5). However, AUR and TAN exhibited concentration-dependently isoform-selective inhibitory activities with IC 50 13.75 and 17.72 μM against PI3Kc, respectively, but no inhibitory activities with IC 50 value >40 μM against PI3Kα, β, and δ isoforms (Table 2 and Figure 5). To our knowledge, it is the first time that AUR and TAN have been reported as PI3K inhibitors.
As a potent PI3Ks inhibitor, the positive PI103 demonstrated excellent inhibition of PI3Ks with IC 50 7.10 nM for PI3Kα, 13.50 nM for PI3Kβ, 85.50 nM for PI3Kc, and 12.50 nM for PI3Kδ as expected.

Molecular Docking Analysis.
e docking results are depicted in Figure 6. As shown in Figure 6(a), LUT bonded to PI3Kc (PDB ID: 1E8W) through six hydrogen bonds, in which the oxygen atom at the pyran (position 4; Figure 7(a)) formed hydrogen bonds with the residue TYR-867 and VAL-882, respectively. e hydroxyl at the benzene ring (position 5; Figure 7(a)) offered hydrogen bonds with the residue VAL-882. e hydroxyl at the benzene ring (position 3; Figure 7(a)) formed hydrogen bonds with the backbone O and the backbone NH of the residue LYS-833. e hydroxyl at the benzene ring (position 4; Figure 7(a)) also formed hydrogen bonds with the residues LYS-833 and ASP-841. In the docking result of TAN with PI3Kc ( Figure 6(b)), the carbonyl group at the pyran (position 4; Figure 7(b)) bonded with the residue VAL-882. e methoxy group at the benzene ring (position 4; Figure 7(b)) formed hydrogen bonds with the residue TYR-867. By the analysis of the binding mode of AUR with PI3Kc ( Figure 6(c)), the carbonyl group at the pyran (position 2; Figure 7(c)) formed hydrogen bond interaction to the residues GLU-880 and VAL-882. e high inhibitory activity of LUT, AUR, and TAN against PI3Kc would be due to these hydrogen bond interactions.

Discussion
Databases, like TCMSP database containing pharmacokinetic properties for natural compounds [18], drug genes Evidence-Based Complementary and Alternative Medicine prediction database like SwissTargetPrediction database [19], gene-related databases like DisGeNET [20] and Gen-eCards [21], and interaction databases like STRING, have been widely used for gathering information for network pharmacology. Besides, appropriate tools like Venny and Cytoscape were adopted. rough compound-target-pathway network, we can directly identify active compounds, targets, and pathways and understand the mechanisms between them in URTIs-associated inflammation. In this study, TAN, LUT, HES, NAR, and AUR were predicted to be the main active compounds related to URTIs-associated inflammation, and MET, MDM2, PIK3CA, IGF1R, and MMP9 were the main URTIs-associated targets. In addition, the KEGG enrichment analysis proved that the PI3K/Akt signalling pathway was the most related pathway. We know that PIK3CA codes PI3Ks, inferring that the active compounds might target at PI3Ks. In vitro PI3Ks inhibition assay demonstrated that the compounds LUT, TAN, and AUR exhibited effective inhibition on PI3Ks. Among them, LUT displayed concentration-dependently inhibitory activities against PI3Ks including Class I α, β, δ, and c isoforms, and AUR and TAN showed concentration-dependently isoformselective inhibitory activities against PI3Kc, which were confirmed by molecular docking with tight docking of these compounds to the PI3Kc in several hydrogen bond interactions.
PI3Ks are the negative regulators involved in regulating anti-inflammatory responses, immune response, compound target pathway Figure 4: Compound-target-pathway network of QFA. e larger nodes represent more important hub nodes. e layout follows the attribute circle with the degree value, and the targets in the outer ring represent more important targets (degree value > median).  Evidence-Based Complementary and Alternative Medicine neurodegenerative diseases, cardiovascular diseases, and tumor [22]. Among three families of PI3Ks, Class I PI3K isoforms (PI3Kα, β, c, and δ) are the most intensively studied compared with the Class II PI3K isoforms (PI3KC2α, C2β, and C2c) and a single Class III PI3K, which consists of a p85 regulatory subunit and a p110 catalytic subunit coded by PIK3CA. Class I PI3K isoforms can be further classified into Class IA isoforms PI3Kα, β, and δ (p110α, p110β, and p110δ) and Class IB isoform PI3Kc (p110c) according to different activating models [23]. As we know, Class IA isoforms PI3Kα and β are ubiquitously expressed, and PI3Kc is expressed in granulocytes, monocytes, and macrophages, whereas the PI3Kδ isoform is found in B-and T-cells. Class IA PI3K isoforms are particularly implicated in human cancers [24]. Class IB isoform PI3Kc is expressed at very low levels under physiological conditions, while it is significantly upregulated after stress. After stress in pathological state, PI3Kc is constitutively enriched in leukocytes. In the airways, PI3Kc behaves as a trigger or a target in a plethora of respiratory diseases, especially in airway inflammation including ALI, pulmonary fibrosis, asthma, and cystic fibrosis [25]. PI3Kc knockout in mouse models of respiratory diseases led to strengthening immunological function and improving airway inflammation, suggesting that PI3Kc is a therapeutic target in inflammatory-driven respiratory diseases. Multiple lines of evidence demonstrated that inhibition of PI3Kc activity attenuated ventilator-induced lung damage, idiopathic pulmonary fibrosis, and asthma [26,27]. In short, PI3Kc has been proved to play a pivotal role in respiratory tract inflammation.
Among the active compounds from QFA, LUT has been described as an anticancer agent, as well as a potent antioxidant and a neuroprotective agent [28], which was used for treatment of lower respiratory infection [29]. PI3Ks assay results demonstrated that LUT acted as a pan-PI3K inhibitor with IC 50 2.49 μM for PI3Kα, 2.95 μM for PI3Kβ, 5.79 μM for PI3Kc, and 1.55 μM for PI3Kδ. It is the first time that LUT has been reported as a pan-PI3K inhibitor at kinase level. PI3Ks were considered to be the molecular target for the antimetastatic effect of LUT, because LUT can directly inhibit PI3K activities and subsequently attenuate phosphorylation of Akt [30]. Moreover, LUT had beneficial effects against lipopolysaccharide-(LPS-) induced ALI, which involved the blockade of MEK/ERK-related and PI3K/Akt-related pathways in neutrophils [31].
TAN, a flavonoid derived from citrus, has been demonstrated to exhibit neuroprotective, antidiabetic, antioxidant, and anti-inflammatory effects [32][33][34]. According to PI3Ks assay results, TAN was a PI3Kc-selective inhibitor, which might be a therapeutic drug to treat URTIs. Other studies reported that TAN could impede human respiratory syncytial virus infections and inflammation [35] and have protective effects against ovalbumin-provoked allergic respiratory asthma in Swiss albino mice [36]. Moreover, there is evidence demonstrated that TAN had antiasthmatic effects partly via regulating PI3K signalling pathway [37].
AUR, a dietary coumarin in citrus fruits, has been found to exert valuable pharmacological properties as anticancer, antibacterial, antiprotozoal, antifungal, anti-inflammatory, and antioxidant agent [38,39]. Furthermore, in vitro inhibition of PI3Ks showed that AUR was a PI3Kc-selective inhibitor which exhibited stronger activity than TAN, suggesting that AUR has a therapeutic effect on URTIsassociated inflammation. Other studies had confirmed that AUR suppressed inflammatory responses in activated RAW 264 macrophages [40,41]. It was also reported that AUR could treat 2 cells diseases such as asthma by inhibiting cell proliferation, decreasing the levels of cytokines (IL-4, IL-10, and IFN-c) and NF-κB, and reducing nitric oxide (NO) production in phytohemagglutinin (PHA) stimulation [42].
According to the KEGG enrichment analysis, TNF signalling pathway was also the related pathway involved in URTIs-associated inflammation. TNF is a key proinflammatory cytokine and has been implicated in many inflammatory lung pathophysiology, including asthma, chronic bronchitis, chronic obstructive pulmonary diseases (COPD), ALI, and ARDS [43]. Several lines of evidence showed that TNF-α level was upregulated in asthmatic patients [44,45]. From the enrichment results of BP, positive regulation of protein serine/threonine kinase (RIPK) had been enriched, which is a kinase that mediates necroptosis in inflammation. RIPK1, a downstream target of the TNF signalling pathway, acts as a molecular switch that induces inflammation and cell survival [46]. When the body is stressed, RIPK1 recruited to TNFR1, and a series of ubiquitination events would activate different signalling pathways [47]. Our study suggested that QFA may also be able to treat URTIs-associated inflammation by interfering with the TNF signalling pathway.
In addition to the above two signalling pathways, NF-κB signalling pathway was also involved in URTIs-associated  Evidence-Based Complementary and Alternative Medicine infection. NF-κB is a primary transcription factor which presents in the cytoplasm with no transcriptional activity in the resting state. After stimulation, phosphorylation of IκB by IκB kinase results in nuclear translocation of NF-κB, which regulates immune, inflammatory, cell proliferation, and apoptosis responses [48]. Li et al. found that QFA extracts attenuated the production of IL-6, IL-1β, and TNFα in LPS-induced RAW 264.7 cells by inhibiting the NF-κB signalling pathway [11]. On the basis of compound-targetpathway network and the KEGG enrichment analysis, the therapeutic effect of QFA on URTIs is related to the inhibition of NF-κB signalling pathways.

Conclusions
By using the method of network pharmacological analyses, five main active compounds (NAR, TAN, LUT, HES, and AUR) of QFA were found to act on many target proteins of URTIs, which reflected the characteristics of multi-compound, multi-target, and multi-action pathways of TCM. e inflammation-related proteins like AKT and PI3Ks played a dominant role in the PPI network and in the compound-target-pathway network, suggesting that the mechanism of QFA for the treatment of URTIs was closely related to PI3K/Akt signalling pathway. e result of in vitro assays and molecular docking results confirmed that TAN, LUT, and AUR, as PI3Kc inhibitors, played a crucial role in respiratory tract inflammation. is paper preliminarily released the material basis and mechanism of QFA in treating URTIs and provided some guidance for the further research on the treatment of URTIs.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
No potential conflicts of interest were reported by the authors.

Authors' Contributions
Shiyi Chen and Wenkang Huang are contributed equally to this work.