The mechanism of Icariside II on NSCLC/COVID-19 based on network pharmacology and molecular docking

Patients with non-small cell lung cancer (NSCLC) are susceptible to coronavirus disease-2019 (COVID-19), but related treatments are limited. Icariside II (IS), a metabolite of plant Epimedin, showed anti-inflammation and immunoregulation effects in various diseases. This study aimed to evaluate the effect and mechanisms of IS on NSCLC/COVID-19. Targets of NSCLC/COVID-19 were defined as the common targets of NSCLC and COVID-19. The clinical characteristics of NSCLC patients were collected from The Cancer Genome Atlas Program (TCGA) database and analyzed by the R package of “survival”, univariate and multivariate Cox proportional hazards regression model. Further, the targets in IS treatment of NSCLC/COVID-19 were defined as the overlapping targets of IS and NSCLC/COVID-19 targets. Gene Ontology (Go) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of the treatment targets found IS might affect the leucocyte migration, inflammation response, and active oxygen species metabolic process, and regulate the IL-17, TNF, and HIF-1 signaling pathway in NSCLC/COVID-19. Protein-protein interaction (PPI) network identified six hub targets of IS in the treatment of NSCLC/COVID-19 including F2, SELE, MMP1, MMP2, AGTR1, and AGTR2. Molecular docking showed above target proteins had a great binding degree to IS. Our finding indicated that IS exerts therapeutic effects in NSCLC patients infected with COVID-19, supporting a further pre-clinical study to validate the related effect and underlying mechanism.


Introduction
The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused a devastating pandemic--Coronavirus disease 2019 (COVID-19)(1). However, no speci c therapy has been developed to overcome this life-threatening epidemic disease yet. Previous study reported an increased incidence of COVID-19 in patients with cancer compared with the general population (2). Cancer patients need to visit the hospital for long-term treatment, which increases the risk of COVID-19 infection.
As any external infection can deteriorate their condition, among patients with cancer and COVID-19, 30day all-cause mortality was higher than patients with the individual disease (3). Lung cancer is the rst leading cause of cancer death (4). More than 350 people will die each day from lung cancer, which is more than breast, prostate, and pancreatic cancers combined and 2.5 times more than CRC, the second leading cause of cancer death (4). Therefore, nding the hub targets and effective therapy in the treatment of NSCLC patients with COVID-19 is of great signi cance in clinical.
Icariside II (IS), also knowns as Baohuoside I, is a metabolite of Icariin from Herba epimedium. IS has been extensively studied for its anti-in ammation and immunoregulation. Recent studies found that IS exhibits a broad spectrum of cytotoxicity against multiple cancer types both in vitro and in vivo (5).
Moreover, studies demonstrated that IS could enhance the anti-tumor effect of the mainstream medication including paclitaxel (6), TRAIL (7), and cisplatin (8). In addition, COVID-19 was considered with cytokine storm syndromes and immunosuppression (9). Thus, we hypothesize that IS might exert potent pharmacological activity in patients with NSCLC combined with COVID-19.
Public databases provide giant resources for data mining. Network Pharmacology is a convenient tool for exploring the potential targets and mechanisms of the therapeutic effect of bioactive components. As non-small cell lung cancer (NSCLC) accounts for about 85% of lung cancer cases, in this study, we collected the NSCLC-related targets from the TCGA database for interaction analysis with COVID-19related targets and performed Clinicopathological analysis of common targets. Then, we explored the mechanism underlying the therapeutic effect of IS on patients of NSCLC combined with COVID-19, based on network pharmacology and molecular docking.

Clinicopathological analysis of NSCLC/COVID-19-related targets
The correlation between NSCLC/COVID-19 targets and survival rates in patients with NSCLC/COVID-19 was analyzed by the R package of "survival". Prognostic analyses were performed using univariate Cox proportional hazards regression. Moreover, different characteristics of diseased targets and patients with NSCLC/COVID-19 were analyzed using a multivariate Cox proportional hazards regression model. Finally, patients were categorized into low-risk and high-risk groups based on the risk score [30]. The optimal cutoff value of risk score was calculated by the R package of maxstat (Maximally selected rank statistics with several P-value approximations version: 0.7-25), setting the minimum number of grouping samples to be greater than 25%, and the maximum number of samples to be grouped less than 75%. The prognostic difference between the two groups was further analyzed by the R package of "survival" and "surv t". The log-rank test method was used to evaluate the signi cance of the prognostic difference between groups.

Enrichment analyses and network visualization of targets that IS in the treatment of NSCLC/COVID-19
The c5.go.bp (cc and mf).v7.4.symbols.gmt and KEGG rest API(https://www.kegg.jp/kegg/rest/keggapi.html) were download from the molecular signatures Database (DOI:10.1093/bioinformatics/btr260, http://www.gsea-msigdb.org/gsea/downloads.jsp) as a background. For gene set functional enrichment analysis, genes were mapped into the background set and enriched using the R package clisterPro ler (version 3.14.3) to obtain the results of gene set enrichment. The minimum gene was set to 5, the maximum gene set to 5000, P-value <0.05, and a FDR<0.1 were considered statistically signi cant. R packages of "GOplot" were used for visualization of gene ontology (GO) biological process (BP) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways of the overlapping gene targets of IS against NSCLC/COVID-19. Finally, a graphical network of drug-target-GO/KEGG pathway-disease of IS in the treatment of NSCLC/COVID-19 was visualized by Cytoscape (3.9.1).

Molecular docking of hub targets and IS
The molecular structure of IS was obtained from the PubChem database (https://pubchem.ncbi.nlm.nih. gov). The protein structure of hub targets was obtained from the PDB database (https://www.rcsb.org/). The Protein Data Bank, Partial Charge (Q), & Atom Type (T) (PDBQT) structure les necessary for virtual screening were created by the AutoDock software (AutoDockTools-1.5.6). Hydrogenation and Gasteiger charge for merging nonpolar hydrogen atoms were performed by the AutoDock software. The visualization was conducted by Pymol software.

Common targets of NSCLC and COVID-19
A total of 2799 NSCLC-related targets were identi ed by using the TCGA database and 488 COVID-19related targets were collected from the database including Genecard (178), OMIM (22), and NCBI (436) ( Figure 2A). Further, sixty-one intersected targets were identi ed between NSCLC and COVID-19 and regarded as the NSCLC/COVID-19 targets ( Figure 2A Supplementary Table 1). Among 61 targets, 22 were up-regulated and 39 were down-regulated in comparing the differential gene expression in tumor and normal ( Figure 2B).
Further, the relationship between risk score, survival status, and expression distribution of the six individual targets in each patient was analyzed. The results demonstrated that greater risk value in patients correlated with higher risk score ( Figure 3C), decreased survival rate of patients ( Figure 3D), and increased expression levels of the three targets-TRPA1, LDHA and HAVCR1 ( Figure 3E). Furthermore, an association of six genes with clinical characteristic in NSCLC were explored ( Table 3). The expression of LDHA, ERG, and risk score were higher in older (>65 years old) than that in younger (≤65 years old) ( Figure 4A C and E). The expression of LDHA, CAT, and risk score were higher in patients with lymph node metastasis (N1-2) than that in patients without lymph node metastasis (N0), indicating that LDHA, CAT, and risk score were related to the number of lymph node metastasis ( Figure 4B D and F).
While no signi cant associations were found between six genes and gender, tumor scope and size (T1-2 and T3-4), and advanced stage (Stage I-II and Stage III-IV).  A total of 565 targets related to IS were predicted from ve drug databases. Eleven intersection targets of IS and NSCLC/COVID-19 were constructed into the PPI network by the STRING database ( Figure 5A, Supplementary Table 2). Further calculation of the topological parameters of the PPI network suggested that the median value of Degree and MNC was two, and the maximum was four. Thus, the hub target screening criteria was set to two-four. Finally, six hub targets were identi ed, including F2, SELE, MMP1, MMP2, AGTR1, and AGTR2 ( Figure 5B, Supplementary Table 3).
Enriched GO and KEGG pathway of targets that IS in the treatment of NSCLC/COVID-19 The 11 targets were submitted for GO and KEGG enrichment analyses, to explore the possible mechanism that IS against NSCLC/COVID-19. Go analysis indicated that IS might affect a series of BPs related to cell (leucocyte) migration, regulation of in ammation response, and regulation of active oxygen species metabolic process ( Figure 6A); IS might affect a series of CCs related to an intrinsic component of plasma member, membrane microdomain, external encapsulating signature, and perinuclear region of cytoplasm ( Figure 6B); IS might affect a series of MFs related to molecular transducer activity, serine hydrolase activity ( Figure 6C). KEGG analysis suggested that IS could regulate the in ammation-related signaling pathway of IL-17 and TNF, HIF-1 signaling pathway, and others including rheumatoid arthritis, renin-angiotensin system, complement, and coagulation cascades, AGE-RAGE signaling pathway in diabetic complications, neuroactive ligand-receptor interaction, serotonergic synapse, phospholipase D signaling pathway and adrenergic signaling in cardiomyocytes ( Figure 6D and E). An interaction network showing hub targets, pharmacological functions, and signaling pathways of IS against NSCLC/COVID-19 was constructed (Figure 8, Supplementary Table 4).

Results of molecular docking of hub targets and IS
Docking simulation studies were carried out to investigate the binding sites of six hub targets with IS.
Discussion COVID-19 pandemic, leading to a gigantic increase in incidence and mortality rates worldwid. It has been reported that patients with lung cancer are more likely to be infected with SARS-COV-2 and have a higher death risk because the poor condition and immunosuppression (17). IS was reported as a potential natural component in anti-in ammation and immunoregulation. To our knowledge, this is the rst study to explore the hub targets and possible signaling pathways underlying the therapeutic effect of IS on NSCLC/COVID-19. We found the hub targets that IS in the treatment of NSCLC/COVID-19 include F2, SELE, MMP1, MMP2, AGTR1, and AGTR2. IS might affect the leucocyte migration, in ammation response, and active oxygen species metabolic process, and regulate the IL-17, TNF, and HIF-1 signaling pathway in NSCLC/COVID-19. Taken together, our work showed that a natural small compound IS might exert therapeutic effects in NSCLC patients infected with COVID-19. Considering the advantages of low cytotoxicity compared to chemotherapeutic drugs, we believe that IS might be a potential agent in the combination strategy for cancer and COVID-19 treatment. At least our work provides evidence for the further preclinical evaluation of IS as a potential natural agent to improve therapies for cancer patients with COVID-19.
In the current study, we rst screened out and identi ed hub genes and 61 intersecting genes of NSCLC combined with COVID-19. The DGE analysis showed 22 upregulated and 39 downregulated genes in patients with NSCLC and/or COVID-19. The DGE-based analysis might be used for determining clinical characteristics in NSCLC patients with COVID-19. Further independent prognostic and survival analyses found some important DGE, including TRPA1, LDHA, KPNA2 HAVCR1, AGTR1, ERVFRD-1, CPA3, SFTPC, ERG, C5AR1, CAT, IL33, C4BPA, TNFRSF13C, TEK, AGER, and SELP. These DGE may function as potent biomarkers for screening and characterizing different stages of NSCLC patients with COVID-19. Taken together, these 61 intersection targets could be potential treatment targets in NSCLC combing with COVID-19.
In the network pharmacology analyses, we identi ed 11 overlapping targets with IS treatment against NSCLC and COVID-19. The DGE analysis identi ed signi cant differences in the expression of AGTR1, LDHA, and TEK. Moreover, NSCLC and COVID-19 showed increased expression of LDHA, and decreased expression of AGTR1 and TEK, along with a lower survival rate. LDHA encodes Lactate dehydrogenase A, which is one of ve isoforms of the lactate dehydrogenase family and plays a crucial role in aerobic glycolysis that is a feature of cancer cells (the Warburg effect) (18). A previous study found phosphorylation-mediated activation of LDHA promotes cancer cell invasion and tumor metastasis (19). Therefore, LDHA is widely regarded as a desirable target for cancer therapeutics. AGTR1 encodes the type 1 receptor which is thought to mediate the major cardiovascular effects of angiotensin II. During SARS coronavirus-2/SARS-CoV-2 infection, it can recognize and internalize the complex formed by secreted ACE2 and SARS-CoV-2 spike protein through DNM2/dynamin 2-dependent endocytosis(20). TEK, encoding receptor tyrosine kinase, belongs to the protein tyrosine kinase Tie2 family. It acts as a cellsurface receptor for ANGPT1, ANGPT2, and ANGPT4 and regulates angiogenesis, endothelial cell survival, proliferation, migration, adhesion, cell spreading, and reorganization of the actin cytoskeleton, but also maintenance of vascular quiescence (21). Receptor tyrosine kinase has anti-in ammatory effects by preventing the leakage of proin ammatory plasma proteins and leukocytes from blood vessels(22).
These ndings indicated that these 11 intersection targets might be potent pharmacological targets of IS in the treatment of NSCLC and COVID-19.
Our GO and KEGG analysis suggested that the mechanism underlying the anti-NSCLC and anti-COVID-19 effects of IS were mediated by regulating leukocyte migration, in ammation response, and active oxygen species metabolic process, and regulate the IL-17, TNF, and HIF-1 signaling pathway in NSCLC/COVID-19. IS, a metabolite of Herba Epimedii has been used for impotency, osteoporosis, and amnestic treatment for thousands of years. IS was reported as a potential anti-in ammatory drug for a series of in ammatory diseases such as atherosclerosis, Alzheimer's disease, depression, osteoarthritis, and asthma(23). Moreover, a recent study demonstrated that IS could potentiate cisplatin-induced apoptosis in non-small cell lung cancer cells (8), overcome TRAIL resistance of melanoma cells (7), and enhance paclitaxel-induced apoptosis in human melanoma A375 cells. Thus, we infer that IS might be a promising agent for patients with NSCLC and COVID-19. Our ndings support the further preclinical study of IS as an anti-tumor and anti-COVID-19 drug.
Moreover, the anti-NSCLC/COVID-19 action of IS could be modulated by hub genes, including F2, SELE, MMP1, MMP2, ARTG1, and ARTG2. Using molecular docking analysis, we identi ed the best binding activities of IS with SER-156 and ILE-128 structures in the hub target F2, indicating that IS can effectively bind to speci c proteins in the novel coronavirus. These ndings suggested that IS may be able to bring the F2 to target the COVID-19. Additionally, we believe that adjuvant supplementation of IS may enhance the therapeutic e cacy of current clinical antiviral agents and immunotherapy to treat the fatal COVID-19, or NSCLC combined with COVID-19.

Conclusion
Network pharmacology and molecular docking indicated IS exerts therapeutic effects in NSCLC patients infected with COVID-19, supporting further pre-clinical studies to validate the related effect and underlying mechanism.

Availability of data and materials
The data used to support the ndings of this study are available from the corresponding author upon reasonable request.