Network Pharmacology and Experimental Validation to Reveal Effects and Mechanisms of Icariin Combined with Nobiletin against Chronic Obstructive Pulmonary Diseases

Background Chronic obstructive pulmonary disease (COPD) is a long-term respiratory disorder marked by restricted airflow and persistent respiratory symptoms. According to previous studies, icariin combined with nobiletin (I&N) significantly ameliorates COPD, but the therapeutic mechanisms remain unclear. Purpose The aim of the study is to investigate the therapeutic mechanisms of I&N against COPD using network pharmacology and experimental validation. Methods The targets of I&N and related genes of COPD were screened and their intersection was selected. Next, the protein-protein interaction (PPI) networks, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed. Further, a COPD rat model was established to validate the effect and mechanisms of I&N. Results 445 potential targets I&N were obtained from SwissTargetPrediction, STITCH 5.0, and PharmMapper databases. 1831 related genes of COPD were obtained from GeneCards, DrugBank, and DisGeNet databases. 189 related genes were screened via matching COPD targets with I&N. 16 highest score targets among 189 targets were obtained according to PPI networks. GO and KEGG pathway enrichment analyses of 16 highest score targets suggested that these key genes of I&N were mostly enriched in the tumor necrosis factor (TNF) pathway, mitogen-activated protein kinase (MAPK) pathway, and phosphatidyl inositol 3-kinase (PI3K)-protein kinase B (AKT) pathway. Therefore, the treatments of I&N for COPD were connected with inflammation-related pathways. In in vivo experiments, the studies indicated that I&N improved the lung function and alleviated the damage of pulmonary histopathology. Moreover, I&N reduced levels of interleukin (IL)-6, IL-1β, and TNF-α in lung tissues of COPD rats and inhibited the activation of the MAPK pathway and PI3K-Akt pathway. Conclusions Icariin combined with nobiletin has therapeutic effects on COPD by inhibiting inflammation. The potential mechanisms of I&N may relate to the MAPK pathway and PI3K-Akt pathway.


Introduction
Chronic obstructive pulmonary disease is the most common disease of the respiratory system with high morbidity and mortality and endangers public health [1]. Lung and systemic infammation and lung injury are the main pathophysiology changes in COPD [2]. Nowadays, various treatment strategies are available for COPD, including bronchodilators and anti-infammatory agents, and bronchodilator therapy is the most common treatment against COPD [2]. However, serious side efects, such as potentially paradoxical bronchospasm, may arise due to adhibition of bronchodilator therapy [3]. Traditional Chinese medicine (TCM) has special superiorities for treating COPD. Bufei Yishen formula (BYF), which is an efective therapeutic strategy for COPD, exerts various positive efects for COPD patients via inhibition of infammation [4]. Icariin and nobiletin, two active ingredients screened from BYF, have been reported to have anti-infammatory, antiapoptosis, and antioxidant efects against several infammatory diseases [5][6][7]. Te efects on improving the lung function and inhibiting the infammatory response of I&N in COPD rats had been proved in previous studies. However, the mechanisms of I&N for treatment of COPD remain unclear and the traditional experimental approaches are difcult to elucidate the mechanisms and key action targets of I&N for COPD.
Network pharmacology is a strategy based on multidirectional pharmacology, system biology, network analysis, and computational biology, which systematically expounds the potential targets and mechanisms of TCM [8]. In this method, the relationship networks of herb, compound, target, pathway, and disease are established, which reveal the molecular basis and forecast the pharmacological mechanisms [9].
In this study, the targets of I&N and related genes of COPD were screened and the ingredients-disease targets network was established. Ten, the potential molecular mechanisms were revealed via gene enrichment analysis and molecular docking. Finally, the COPD rat model was established to verify therapeutic efects and potential pathway of I&N against COPD ( Figure 1). Terefore, the primary goals of this study were (1) to screen related genes of COPD and the potential targets of I&N; (2) to dissect the underlying mechanisms of I&N for COPD using network pharmacology; and (3) to validate anti-infammatory efects and the potential pathway of I&N for treatment of COPD.

PPI Network Construction.
First, we intersected the obtained components targets with the genes associated with COPD and obtained a Venn diagram of the intersected gene symbols. Ten, a PPI network was built using STRING [17] and Cytoscape 3.8.2. To screen the key targets, the topological characteristics were analyzed of the PPI network. First, the gene symbols were chosen by the degree score. Next, the betweenness centrality (BC), closeness centrality (CC), degree, and average shortest path length (ASPL) were calculated by Cytoscape to indicate the potential targets.

Enrichment of GO and KEGG Pathways.
Te GO and KEGG pathways enrichments of the topological potential targets were analyzed in DAVID 6.8 [18]. Te p value <0.05 was set as a signifcant diference for KEGG pathway analysis.

Molecular
Docking. Te 3D structures of icariin and nobiletin were acquired from PubChem and were transformed from their original constructions into PDB formats using Open Babel 3.1.1. From RCSB Protein Data Bank, the X-ray crystal structures of key proteins were obtained [19]. Seven protein targets were studied: AKT1 (PDB ID: 2UZR), TNF (PDB ID: 7KP9), VEGFA, (PDB ID: 7LL8), EGFR, (PDB ID: 5Y9T), JUN, (PDB ID: 5T01), MMP9, (PDB ID: 1L6J), and SRC, (PDB ID: 2BDF). Te water molecules were deleted and hydrogen atoms were added in optimizer of structures using AutoDock Tool 1.5.6. Ten, the receptor proteins were docked with ligand molecules via AutoDock. All of options were default setting for docking run. Finally, the molecular docking results were visualized by PyMoL 2.2.3, which acquire the highest scores. (Heilongjiang, China). Te rat ELISA kits of IL-6 (Cat.No.550319) were purchased from BD Biosciences (California, America). Te rat ELISA kits of IL-1β (E-EL-R0012c) and TNF-α (E-EL-R2856c) were purchased from Elabscience Biotechnology Co., Ltd (Wuhan, China). Te antibodies for rat of PI3K (GTX55747, Gene Text) and P-AKT (GTX128414, Gene Text) were obtained from Gene Tex, Inc (North America). Te antibody for rat of P-p38 (4511, CST) was obtained from Cell Signaling Technology (Shanghai, China). Te antibody for rat of GAPDH (10494-2 Evidence-Based Complementary and Alternative Medicine 1-AP, Proteintech) was purchased from Proteintech (Wuhan, China).

Establishing the COPD Rat Model.
A COPD rat model was performed in terms of previous studies [20]. SD rats were randomly classifed to 4 groups: control group, COPD model group, I&N group, and doxofylline group. Te COPD rat model was created via exposure to cigarette smoke (CSE) and Klebsiella pneumoniae infection. Specifcally, the rats were exposed to CSE (3000 ± 500 ppm) for 40 minutes twice daily for 8 weeks and to Klebsiella pneumoniae (6 × 10 8 CFU/ ml, 0.1 ml) for 5 days once for 8 weeks. Te procedures of this study were approved by the Experimental Animal Care and Ethics Committees of the First Afliated Hospital of Henan University of Chinese Medicine, and the ethical review approval number is YFYDW2019031.

Drug and Treatment.
From week 9, the I&N group rats were given I&N at 2.12 mg/kg/d (the ratio of icariin to nobiletin was 12.5 : 1). Te doxofylline is a newer generation xanthine, which is a kind of efective bronchodilator recommended by Global Initiative for Chronic Obstructive Lung Disease (GOLD) [2]. Te doxofylline has benefcial efects with both bronchodilating and anti-infammatory activities in COPD 1. So, we chose doxofylline as the control drug. Te doxofylline group rats were given doxofylline at 36 mg/kg/d. Te dosages of these drugs were calculated according the following formula (D: dose; K: body shape index, K � A/W 2/3 , A: surface area in m 2 , W: weight in kg):

Evidence-Based Complementary and Alternative Medicine
At week 17, 4 group rats were sacrifced after intraperitoneal injection of 2% pentobarbital sodium at 40 mg/ kg.

Lung Function Measurement and Lung Tissue
Histopathology. Lung function was detected for all group rats every four weeks from 0 week to 16 th week via the tidal volume (TV), peak expiratory fow (PEF), and 50% tidal volume expiratory fow (EF50) by unrestrained pulmonary function testing plethysmographs (Buxco Inc., Wilmington, NC, USA).
Te lung tissues were soaked in 4% paraformaldehyde solution. Next, the tissues were cut and embedded in parafn and made slices. Ten, the lung tissues slices were stained with hematoxylin and eosin and were observed by a light microscope (Olympus, Tokyo, Japan). Te mean linear intercept (MLI) and mean alveolar numbers (MAN) were considered as the degree of alveolar damage. Under microscopy (×200), 6 visual felds were taken in each slice, and the alveolar number and the linear intercept in a fxed area of visual feld were measured. MAN (/mm 2 ) � Na/A. Na is the number of pulmonary alveoli in each visual feld. A is the area of the visual feld. Ten, we made a cross (+) under the visual feld and counted the number of alveolar septaon the cross. MLI (μm) � L/Ns. Ns is the number of alveolar septa. L is total length of the cross.

ELISA.
Te lung tissue was homogenized in PBS solution and centrifuged to collect the supernatant. Te secretion of TNF-α, IL-1β, and IL-6 in a lung tissue homogenate was measured using ELISA kits, according to the manufacturer instructions. Te dilution ratio of the lung tissue homogenate was determined according to the standard curve. Samples were incubated with antibodies in 96well plates. Te OD value was detected by a microplate reader (Termo Fisher Scientifc 1500, Vantaa, Helsinki, Finland), and the concentration was calculated according to the standard curve.

Western Blotting Assay.
Te lung tissues were lysed with RIPA bufer in ice to obtain protein samples. Te concentrations of lung tissue protein samples were measured using BCA kits, and the lung tissue protein samples were adjusted to equal concentrations. Te lung tissue protein samples with equal concentrations in each group were divided by SDS-PAGE electrophoresis and metastasized to PVDF membranes. 5% skim milk was used to block the PVDF. Next, membranes were incubated with their primary antibodies, including GAPDH (1 : 5000), P-p38 (1 : 1000), P13K (1 : 1000), and P-AKT (1 : 1000), and secondary antibodies (1 : 5000). Te membranes were visualized using the Bio-Rad Imaging System (Pierce, USA).

Statistical Analysis.
Te experimental data were analyzed by SPSS v21.0. A comparison among groups was performed by one-way analysis of variance with an appropriate post-hoc test. If the variances were homogeneous, the LSD method was performed. If the variances were inconsistent, Dunnett's T3 test was performed. Te mean ± SD were used as the data present presentation. A p value of <0.05 was set for a statistically signifcant diference.

Screening Targets of Components and COPD. From
PubChem, the 2D structures of icariin and nobiletin were downloaded (Figure 2(a)). 445 genes were obtained as potential targets of icariin and nobiletin from Swis-sTargetPrediction database, STITCH database, and PharmMapper database. Ten, 1,831 related genes of COPD were obtained from DisGeNET database, GeneCards database (score >15.0), and DrugBank database. Matching COPD targets with icariin and nobiletin targets, 189 genes ( Figure 2(b)) were chosen as related genes of I&N against COPD for constructing the component-target (C-T) network (Figure 2(c)). Te C-T network was built by Cytoscape software. According to the C-T network, 120 potential targets were common targets of icariin and nobiletin. 59 potential targets were unique targets of icariin and 10 potential targets of nobiletin.

Protein-Protein Interaction (PPI) Network Analysis.
All of 189 potential therapeutic targets were submitted to STRING database, and they were submitted to Cyto-Scape3.8.2 for constructing and analyzing the PPI network ( Figure 3). Te PPI network consisted of 189 nodes and 2809 edges and the average degree was 23. Ten, the targets with degree higher than double average degree were selected and 41 targets were screened for further analysis. Next, the mean value of BC, CC, ASPL, and degree of 41 targets were calculated using the Analyze Network tool of Cytoscape3.8.2. Te targets with values of BC, CC, and degree higher than the mean value of BC, CC, and degree (BC > 0.0092, CC > 0.5875, degree > 60), and value of ASPL lower than mean value of ASPL (ASPL < 1.7021), were selected as key targets. Finally, 16 targets were screened out, including TNF, AKT1, VEGFA, EGFR, JUN, SRC, MMP9, CASP3, MYC, IGF1, HSP90AA1, HRAS, ESR1, PTGS2, PPARG, and MAPK1 ( Figure 3).

Enrichment Analysis of the GO and KEGG Pathways.
Te DAVID 6.8 database was used to perform GO and KEGG analyses on 16 important targets. Positive regulation of transcription from the RNA polymerase II promoter, negative regulation of the apoptotic process, and positive regulation of transcription, DNA-templated, were mostly enriched in BP enrichment analysis; nucleus, cytoplasm, and cytosol were mostly enriched in CC enrichment analysis; protein binding, identical protein binding, and enzyme binding were mostly enriched in MF analysis. (Figure 4(a)). Te results of KEGG analysis indicated that the regulatory pathway included TNF, MAPK, and PI3K-Akt pathways (Figure 4(b)). Tese results suggested that I&N may exert inhibition efects of infammation in COPD by regulating the TNF, MAPK, and PI3K-Akt pathways. Te component-      Evidence-Based Complementary and Alternative Medicine target-pathway network was built by Cytoscape software (Figure 4(c)).

Molecular Docking.
To clarify the potential interaction between two components and the key proteins, molecular docking was performed to reveal the possible binding mode between the 7 highest scoring proteins, including TNF, AKT1, VEGFA, EGFR, JUN, SRC, and MMP9 (Figure 4(d)), and two components. Te binding energy was considered as an important factor for constituents screening (Table 1)

Efects of I&N on the Lung Function in COPD Rats.
To verify treatment of I&N of COPD, we established the COPD model through co-treatment with CSE and Klebsiella pneumoniae in rats. As described in Figure 6, compared with the control group, the TV, PEF, and EF50 in lung functions descended signifcantly in COPD rats (P < 0.05), and I&N and doxofylline increased the TV, PEF, and EF50 in rats (P < 0.05).

Efects of I&N on Lung Tissue Histopathology in COPD
Rats. Lung tissue histopathology analysis indicates that I&N reduced alveolar damage and airway wall thickness (Figure 7(a)). Quantitative analysis of lung tissue histopathology showed that (Figure 7(b)), compared to the control group, MAN was decreased and MLI was increased in COPD rats (P < 0.05); I&N and doxofylline increased MAN and decreased MLI (P < 0.05); and I&N efectively relieved the thickened airway wall in COPD rats (P < 0.05).

Efect of I&N on the Infammatory Response and
Infammation-Related Pathway. In the COPD rats, the mRNA levels and protein secretion of infammatory factor in lung tissues were signifcantly increased, including IL-6, IL-1β, and TNF-α, and these were decreased with treatment of I&N (P < 0.05) (Figures 8(a) and 8(b)). As in Figures 8(c)   Evidence-Based Complementary and Alternative Medicine 7 and 8(d), the expression of PI3K, P-AKT, and P-p38 of lung tissues were signifcantly increased in the model group, and I&N decreased the expression of PI3K and phosphorylation of P-AKT and P-p38 of lung tissues in COPD rats (P < 0.05). Tese results suggested that I&N inhibit infammatory responses in COPD rats via regulating the PI3K-AKT and MAPK pathways.

Discussion
It has been verifed that TCM has positive therapeutic efects on COPD. BYF, a TCM therapeutic strategy for COPD, has demonstrated that it can inhibit secretion of infammatory cytokine, recover protease-antiprotease imbalance, and reduce collagen deposition [22]. Due to the complicacy of TCM ingredients, it is difcult to explore potential therapeutic mechanisms of BYF. Terefore, fve critical active ingredients of BYF were screened out and integrated into efective-component compatibility of Bufei Yishen formula (ECC-BYF), including icariin, nobiletin, astragaloside IV, 20-S-ginsenoside Rh1, and paeonol. It has been verifed the treatment efects of ECC-BYF for COPD on improving the pulmonary function and reducing pathological damage and the infammatory cytokine levels in lung tissues in COPD rats [23]. Icariin and nobiletin are two main active ingredients of ECC-BYF. In a previous study, we found the efects of I&N on improving the lung function, reducing pathological damage, and inhibiting infammatory response in COPD rats. However, the therapeutic mechanisms of I&N for COPD remain unclear. In this study, we devote to reveal the treatments and mechanisms of I&N against COPD. Terefore, we integrated network pharmacology and experiment verifcation to systematically evaluate the potential pharmacological mechanisms of I&N for COPD. First, we applied network pharmacology to screen the possible targets of I&N against COPD. 189 targets of I&N in COPD were obtained from 6 databases, and those with BC > 0.0092, CC > 0.5875, degree > 60, and ASPL < 1.7021 were considered as key targets. 16 key targets were screened out from the 189 targets via PPI network analysis, including TNF, AKT, and MAPK1. Tese key targets were signifcantly related to infammation. Furthermore, the 16 key targets were mostly enriched infammation-related pathway according GO analysis and KEGG analysis, such as TNF, PI3K-AKT, and MAPK signaling pathways. Te result suggested I&N may inhibit the infammatory response in COPD via these proteins and pathways. Ten, molecular docking of I&N and these proteins was performed to verify the possibility of interaction, and these proteins, including AKT, TNF, EGFR, and MMP9, had strong binding energy with I&N.
Infammation is a key pathological reaction for the development of COPD [24]. Te main infammatory cells in COPD involve neutrophils, macrophages, and lymphocytes in the lung tissue and airway [25]. Te infammatory mediators and destructive enzymes from infammatory cells are related to the structural damage of the airway and lung tissue in COPD [26]. For instance, neutrophils in COPD patients and COPD model rats are recruited to the lung and airway and secrete various serine proteases, including myeloperoxidase (MPO), matrix metalloproteinase (MMP), and neutrophil elastase (NE), all of which are related to destruction of the alveolar airway and cause emphysema [27]. PI3K, a kind of lipid kinases, induced the phosphorylation of AKT to regulate cell survival, growth, multiplication, and death in response to extracellular signals. Based on previous studies, the infammatory efcacy of the PI3K-AKTsignaling pathway in COPD. Te concentrations of TNF-α and IL-6 in both the bronchoalveolar lavage fuid (BALF) and serum are decreased via restraining the activation of PI3K-AKT signaling in COPD model rats [28]. Macrolide reduces lung and systemic infammation of COPD patients by regulating the PI3K-AKTN pathway [29]. Te family of MAPKs, including p38, ERK, and JNK, is considered as a signifcant role in the infammatory process [30]. Te MAPK signaling pathway regulates COPD-related characteristics such as chronic infammation and cytokine expression. Te levels of phosphorylation of ERK, p38, and JNK in RAW 264.7 cells stimulated by CSE are much higher, indicating that MAPK signaling was activated in macrophages. Treatment with a MAPK signaling inhibitor also successfully inhibited the TNF-α, IL-1β, and HO-1 overexpression following CSE [31].
Moreover, PI3K-AKT and MAPK signaling pathways are considered as the major pathways, which observably upregulate the MUC5AC expression with the elevated phosphorylation level [32]. MUC5AC, a major secreted mucin which is closely connected with the viscoelasticity of sputum, endangers mucociliary functions and decreases mucus clearance because of secretion excessive, and leads to aggravated lung infection [33]. According to previous studies, the secretion of MUC5AC was downregulated via the inhibition of PI3K-AKT signaling pathway phosphorylation [34]. It has been reported that icariin inhibits CSE-induced infammation, ROS production, and airway remodeling via mitigating glucocorticoids resistance in CSE-exposed BEAS-2B cells [35]. In addition, nobiletin exhibited protective efects in decreasing the production of TNF-α, IL-6 via restraining activation of NF-κB signaling in the LPSinduced acute lung injury mice model and LPS-stimulated A549 cells [36]. We had validated the anti-infammatory efect of I&N against COPD in in vivo experiment. Te mRNA and protein expression levels of IL-6, IL-1β, and TNF-α in lung tissues of COPD model rats were signifcantly increased and were decreased by I&N and doxofylline. On the other hand, the decline of lung function and emphysema is a common symptom during the development of COPD [37]. In in vivo experiments, the lung function and alveolar damage were signifcantly improved by treatment of I&N and doxofylline compared to the model group. Furthermore, the expression levels of PI3K and phosphorylation levels of P-AKT and P-p38 in lung tissues were signifcantly decreased after the treatment of I&N and doxofylline compared to the model group. Tese results confrm the inhibition infammatory response efects of I&N in by decreasing the expression levels of infammatory cytokines. Moreover, doxofylline can improve the lung function and the expression of infammatory factors in COPD rats. Te therapeutic efects of I&N were consistent with those of doxofylline in improving symptoms and inhibiting infammation. Moreover, the potential mechanism may be related to suppress the phosphorylation of the PI3K-AKT and MAPK pathway in COPD.

Conclusion
In our research, the therapeutic efcacy and mechanisms of I&N for COPD are verifed via the method integrating network pharmacology and experiment validation. 16 key targets of I&N against COPD were screened, including TNF, AKT1, and MAPK1. According KEGG pathway analysis, the activation of the MPAK and PI3K-AKT pathways was a signifcant mechanism of I&N against COPD. In in vivo experiments, the lung function, pathological damage of lung tissues, and secretion of IL-6, IL-1β, and TNF-α were improved by treatment of I&N in COPD rats. Furthermore, the levels of PI3K, P-AKT, and P-p38 were reduced by I&N. In conclusion, I&N have signifcant anti-infammation efects for COPD via the restraining activation of PI3K-AKT and MPAK pathways. However, the complex mechanisms of I&N for treatment of COPD require further exploring.

Data Availability
Te data used to support the fndings of this study can be obtained from the corresponding author according to the rules.

Conflicts of Interest
Te authors declare that they have no conficts of interest.