Potential Mechanism Prediction of Herbal Medicine for Pulmonary Fibrosis Associated with SARS-CoV-2 Infection Based on Network Analysis and Molecular Docking

Background: Coronavirus Disease 2019 (COVID-19) is still a relevant global problem. Although some patients have recovered from COVID-19, the sequalae to the SARS-CoV-2 infection may include pulmonary fibrosis, which may contribute to considerable economic burden and health-care challenges. Convalescent Chinese Prescription (CCP) has been widely used during the COVID-19 recovery period for patients who were at high risk of pulmonary fibrosis and is recommended by the Diagnosis and Treatment Protocol for COVID-19 (Trial Version sixth, seventh). However, its underlying mechanism is still unclear. Methods: In this study, an integrated pharmacology approach was implemented, which involved evaluation of absorption, distribution, metabolism and excretion of CCP, data mining of the disease targets, protein-protein interaction (PPI) network construction, and analysis, enrichment analysis, and molecular docking simulation, to predict the bioactive components, potential targets, and molecular mechanism of CCP for pulmonary fibrosis associated with SARS-CoV-2 infection. Results: The active compound of CCP and the candidate targets, including pulmonary fibrosis targets, were obtained through database mining. The Drug-Disease network was constructed. Sixty-five key targets were identified by topological analysis. The findings of Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotation suggested that the VEGF, Toll-like 4 receptor, MAPK signaling pathway, and TGF-β1 signaling pathways may be involved in pulmonary fibrosis. In the molecular docking analyses, VEGF, TNF-α, IL-6, MMP9 exhibited good binding activity. Findings from our study indicated that CCP could inhibit the expression of VEGF, TNF-α, IL-6, MMP9, TGF-β1 via the VEGF, Toll-like 4 receptor, MAPK, and TGF-β1 signaling pathways. Conclusion: Potential mechanisms involved in CCP treatment for COVID-19 pulmonary fibrosis associated with SARS-CoV-2 infection involves multiple components and multiple target points as well as multiple pathways. These findings may offer a profile for further investigations of the anti-fibrotic mechanism of CCP.


INTRODUCTION
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a newly discovered coronavirus responsible for COVID-19, which causes atypical pneumonia progressing to acute respiratory distress syndrome (ARDS) and acute lung injury (Marini and Gattinoni, 2020). After SARS-CoV-2 infection, patients who have experienced and survived the COVID-19 outbreak may face a greater risk of developing pulmonary fibrosis (PF), which is a chronic, severe, and progressive interstitial lung disease (George et al., 2020). A meta-analysis demonstrated that there was a clear association between the development of PF and respiratory viral infection (Hutchinson et al., 2015). A well-known mechanism is that SARS-CoV-2 invades host cells and interacts with ACE2, which is highly expressed in pneumocytes type II cells and is directly involved in the initiation and progression of inflammation and fibrosis (Ksiazek et al., 2003;Li et al., 2003;Xu et al., 2020). Importantly, PF not only forms as a sequelae to chronic inflammation but is also is genetically influenced by an age-related fibroproliferative process such as idiopathic PF. Further, PF is a well-recognized sequela of ARDS (Burnham et al., 2014). Existing data show that about 40% of COVID-19 patients may develop ARDS, which accounts for a high percentage of COVID-19 patients (Wu et al., 2020). Although in most patients, the virus in patients recovering from COVID-19 has been eradicated, this does not prevent the development of PF. Given these observations, PF after recovery from COVID-19 may result in a substantial medical burden and health-care challenges. Therefore, preventing PF in patients recovering from SARS-CoV-2 infection is an urgent issue that needs to be addressed.
Currently, there are few anti-fibrotic drugs available with clinically positive results, and a limited number of such agents are under investigation (Canestaro et al., 2016). Although the antifibrotic medicines, such as pirfenidone and nintedanib approved by FDA, could delay the decline in lung function, these drugs may not improve the quality of life (QoL) or reduce early mortality (Rochwerg et al., 2016;Brown et al., 2020). More importantly, these drugs may not be prescribed for severe or critical cases with COVID-19 on mechanical ventilation due to oral use only. Moreover, pirfenidone and nintedanib are associated with a high incidence of drug-related side effects such as abnormal liver function and uncomfortable clinical manifestations such as gastrointestinal disturbances and skin reactions (King et al., 2014;Kim and Keating, 2015;Noble et al., 2016;Hanta et al., 2019), which have restricted their clinical application. Thus, the development of an effective therapeutic strategy for PF is urgent.
Traditional Chinese medicines (TCM) have been widely used to treat lung diseases. Recently, several meta-analyses have shown that TCM exerted positive effects on PF, such as delaying the decline of pulmonary function and improving the QoL, with a good safety profile Ji et al., 2020). In addition, several pharmacological studies are investigating TCM as an anti-PF treatment. Some studies have revealed that TCM could effectively resist oxidative lesion, histopathological damage (Sun et al., 2018), and reverse extracellular matrix (ECM) as well as Lox2 proliferation by modulating MAPK activation and suppressing the TGF-β/Smad pathway (Tao et al., 2017). A convalescent Chinese prescription (CCP) was widely used for patients with COVID-19 who were in the recovery period and were at high-risk of pulmonary fibrosis, as recommended by the Diagnosis and Treatment Protocol for COVID-19 (Trial Version sixth, seventh) published by the National Health Commission of the People's Republic of China (National Health Commission of the People's Republic of China, 2020a; National Health Commission of the People's Republic of China, 2020b).
Although clear clinical benefits exist, very little has been elucidated about the potential molecular mechanism involved. Network pharmacology is a branch of pharmacology that uses network methods to analyze the synergistic relationship among drugs and diseases and targets via "multi-component, multitarget, multi-pathway" analyses, and can build a multidimensional network model of "drug-component-target-disease" to explore the relationship between drugs and diseases (Missiuro et al., 2009;von Mering et al., 2003). This study was based on network pharmacology and systematically analyzed the effective ingredients, potential targets, pathways, and biological processes of CCP used during the recovery period of COVID-19. The study screened the main active ingredients of CCP via a molecular docking approach to explore the potential molecular mechanisms of action involved in CCP interference with PF associated with SARS-CoV-2 infection. The flowchart of the whole study design is illustrated in Figure 1.

Active Components Database
CCP contains 18 types of Chinese herbal medicine (

Screening of Active Ingredients
The critical parameters of oral bioavailability (OB), drug-likeness (DL), and drug half-life (HL) were used to screen the active components of CCP. OB is an essential indicator for objective Frontiers in Pharmacology | www.frontiersin.org April 2021 | Volume 12 | Article 602218 evaluation of the internal quality of drugs. OB defines the percentage of an orally administered dose of unchanged drug that reaches the systemic circulation and represents the convergence of the absorption, distribution, metabolism, and excretion (ADME) process. High OB is often a key indicator to determine the "drug-like" properties of bioactive molecules as therapeutic agents. Molecules with OB >30% were considered to have good OB in the present study (Xu et al., 2012). DL is a qualitative concept used in drug design to estimate the drug-like properties of a prospective compound and helps to optimize pharmacokinetics and pharmaceutical properties, such as solubility and chemical stability. The "drug-like" level of the compounds was set at 0.18, which is used as a selection criterion for the "drug-like" compounds for traditional Chinese herbs (Tao et al., 2013); thus, ingredients with DL > 0.18 were selected. The HL (t1/2) means that the time it takes to reduce the number of compounds in the body by half, is arguably the most important property of an active ingredient as it dictates the timescale over which the compound may elicit therapeutic activity . HL values > 4 h were selected.

Drug and Disease Target Fishing
We screened for potential targets of the herbs constituting CCP in the TCMSP database. If there was no corresponding drug target in the database, we determined the molecular structural formula based on available chemical formulas from the literature and predicted their potential targets based on the spatial structure of the molecular structure formula in the Swiss-target prediction.
The final potential targets of these herbs were obtained by screening the corresponding drug targets for constituents and FIGURE 1 | Schematic diagram of the integrated pharmacology strategy approach that combines quantitative analysis of components, network analysis, and molecular docking to investigate the mechanisms of Convalescent Chinese prescription (CCP) treatment against pulmonary fibrosis. removing duplicates. The qualitative targets were matched to the UniProt database for normalization (Missiuro et al., 2009). In this study, PF was considered a phenotype in a convalescent patient with SARS-CoV-2 infection. Therefore, the targets related to pulmonary fibrosis were explored based on the OMIM database, drug bank database, and the DisGeNET database. We merged all queried targets and eliminated duplicated results. Finally, standardized names were implemented via the UniProt database.

Network Construction and Analysis
To further explore the mechanisms of CCP's treatment effects on PF of patients with COVID-19 during the rehabilitation stage, we established a network drug-disease map to show the association between the active components in the CCP and its potential targets using Cytoscape v3.7.1. The components and targets were represented by triangles and circles, respectively, and the interaction between the two was shown by a connecting line.
Overlapping portions between drug targets and disease targets are represented by Wayne's diagram. To explain the interaction between target proteins, overlapping target proteins between CCP and PF were uploaded to STRING to obtain information on protein-protein interaction (PPI) . We selected a medium confidence data threshold of >0.4. The obtained protein-protein interaction data were submitted to Cytoscape 3.7.1 to build a PPI network. Additionally, the Significant Gene Ontology (GO) Pathway and the Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway were screened by the DAVID database.

Molecular Docking
The above-mentioned PPI analysis generated potential hub genes active in PF treatment. Binding activity between the active drug components and the hub genes were evaluated by flexible molecular docking using Surflex-Dock software (Jain, 2007). Surflex-Dock uses an idealized active site ligand called a prototype molecule as a target for generating a hypothetical conformation of a molecule or a molecular fragment (Jain, 2007). These hypothetical conformations are all scored by the Hammerhead scoring function, which also serves as the objective function for the local optimization of the conformation (Welch et al., 1996). Through the crossover process, a large number of conformations are assembled from the complete molecule to achieve flexible docking. The crystal structure of the target proteins was obtained from the Protein Data Bank (Berman et al., 2003). The structural formula (MOL2 format) of the compounds were available at the TCMSP database and PubChem. If the structural formula was not available, we would manually draw the molecular structure with ChemDraw software. The protein targets were processed by removing water, adding hydrogen, and extracting the ligand structure accordingly and finally, Surflex-Dock v.2.1 was run to perform molecular docking (Lill and Danielson, 2011;Spitzer and Jain, 2012;Yuan et al., 2016). In order to examine the stability of molecular Docking, we also perform another molecular docking (dockthor, https://dockthor.lncc.br/v2/). A similar result occurs if molecular docking is relatively precise.

Active Component Screening
The active components of CCP were retrieved from the TCMSP database based on three parameters (OB>30%; DL > 0.18; HL > 4). Ultimately, 308 related components were identified as active ingredients in the CCP. Milkvetch root (HQ, 16 ingredients), Tangshen (Guoqiang et al., 2018;Wu et al., 2018). The potential compounds of the CCP formulation and the respective ADME parameters are shown in detail in Supplementary Material S1.

Target Fishing for Drug Components and Disease and Establishing the Drug-Disease Target Network
Targets from 14 Chinese herbs were available in the TCMSP database. Targets from Leech (SZ) were available from a previous report (Guoqiang et al., 2018). Targets from Ground beetle (TBC) were available in Swiss-target Prediction software according to its components (Wu et al., 2018). Targets from Dwarf lilyturf tuber (MD) and Hawthorn fruit (CSZ) were available in the Chemistry Database. We matched the components of herbals with the corresponding targets. We obtained 307 targets of HQ, 102  The intersection (65 common targets) between drug targets and disease targets is shown in Figure 2, and the details of the shared targets are shown in Table 2. Subsequently, we mapped the components of CCP based on these 65 targets to construct the drug (components)-disease (PF targets) network shown in Figure 3. Next, we submitted the 65 targets to the string tool to generate the PPI network. The PPI network was also visualized using Cytoscape 3.7.1 software. As shown in Figure 4A, 65 nodes and 683 edges were identified in the PPI network (Network Properties: Degree 20, Betweenness 21.65795, Closeness 40.83333). Subsequently, a topological analysis of the PPI network was implemented using network properties values greater than the median values (Degree 34, Betweenness 65.59, Closeness 50.25), as shown in Figure 4B. The identified targets (AKT1, TNF, IL6, TP53, VEGFA, IL1B, MMP9, EGFR, CCL2, PTGS2, STAT3, EGF, SRC, MMP2, FOS, CAT, HMOX1, ICAM1, MMP1, TGFB1) represented potential key targets for the therapeutic effects of CCP ( Figure 4C).

Enrichment Analysis
Using the DAVID database, GO enrichment analysis yielded GO entries (p < 0.05) comprising 336 biological processes (BP), 30 cellular components (CC), and 27 molecular functions (MF). The top 20 entries were selected from BP, CC, and MF, respectively, in order of -lgP value ( Figure 5). In the BP, the primary target in the extracellular region was the cytosol; for MF, the targets mainly involved enzyme binding, protein binding, and identity protein binding, and cytokine activity.
In total, 93 terms (p < 0.05) were obtained from the KEGG pathway enrichment analysis using DAVID data. The first 20 entries were selected according to the-lgP value to draw a bubble diagram ( Figure 6)

Molecular Docking
The binding ability and of herbal components to core protein targets were validated by molecular simulations. Using molecular docking by Surflex-Dock modeling, a docking score greater than 3 was considered as a stable compound binding to the protein. In this respect, the top 20 nodes were selected from the drug-disease network (Figure 3 Figure 7, quercetin, kaempferol, and luteolin exhibited high binding activity to targets associated with PF. for example IL-6 (score 3.0236, 3.6316, 3.7055, respectively), TNF-α (score 3.2116, 3.9889, 5.9409, respectively), VEGF (score 3.0175, 3.844, 3.1564, respectively), MMP9 (score 5.7384, 3.079, 5.9618, respectively). Detailed blinding scores were shown in the Heat map in Figure 7 and in the Supplementary

DISCUSSION
CCP has been described as a suitable treatment for the recovery phase of patients with COVID-19. In this study, a total of 116 active ingredients and 583 action targets of CCP were screened based on the network pharmacology approach. By further screening, 72 core compounds and 26 key action targets were obtained. Network pharmacology analysis embodies the holistic and correlative characteristics of the combined action of multiple components and targets of traditional Chinese medicine. According to the topological property analysis of the "drug-disease" network, the top 5 key compounds in CCP are quercetin, kaempferol, luteolin, TBC2, and TBC4. In addition, other the components were collected through multiple databases such as China Knowledge, PubMed, and BATMAN-TCM. The chemical constituents (TBC2, TBC4, TBC5) are not included in the TCMSP platform. The targets of these components need to be further studied. Quercetin is known to possess marked antioxidative, anti-inflammatory, and antifibrotic capacities. Dietary quercetin supplementation also decreases chronic systemic inflammation (Stewart et al., 2008). Quercetin can reduce the expression of transforming growth factor-β1 (TGF-β1), α-smooth muscle actin (α-SMA), and tumor necrosis factor-α (TNF-α), inhibit alveolar cell apoptosis and reduce lung tissue inflammation and fibrosis injury in rats, which can effectively improve lung fibrosis Zhang et al., 2018;Wei et al., 2019). Kaempferol, a polyphenol with potent antioxidant activity, is an ester of caffeic acid and quinic acid (Nardini et al., 2002). Luteolin exhibits various pharmacological activities, including antioxidative, antiviral, antibacterial, anti-inflammatory, anti-endotoxin, antitumor, and liver-protective effects (Rohit and Mohan Rao, 2013;Tajik et al., 2017;Wei et al., 2017). According to the results of the PPI network analysis and the topological property analysis of the "Drug-disease" network, the targets of action of CCP were the cytokine IL-6, members of the MAPK family, and PTGSE or prostaglandin G/H synthase. The results of GO functional enrichment analysis revealed that CCP components were mainly involved in the cell communication, endogenous stimulus regulation, apoptosis, programmed cell death, steroid hormone stimulus, signal transduction, regulation of catalytic activity, wounding regulation of cell Protein-protein interaction (PPI) networks of shared targets between Convalescent Chinese prescription (CCP) and pulmonary fibrosis were analyzed by STRING 11.0. (B). The most significant module identified by the topology selection (degree centrality >34, betweenness centrality >65.59, closeness centrality >50.2). (C). The core 20 targets (hub targets) in the PPI network ranked by degree centrality using the cytoHubba plug-in.
FIGURE 5 | Enrichment analysis of the potential targets of Convalescent Chinese prescription (CCP) against pulmonary fibrosis by R software 3.4.2 for the Gene Ontology database. Top 20 biological process (BP) terms, cellular component (CC) terms, and molecular function (MF) terms are shown as green bars, orange bars, and purple bars, respectively, according to "p-value (<0.05), Bonferroni correction." proliferation, regulation of chemokine production and multicellular organismal macromolecule metabolic process. The KEGG pathway enrichment mainly involved Toll-like 4 receptor signaling and immune response pathways. Additionally, the NOD-like receptor signaling pathway and the Graft-versus-host disease pathway involved the IL-6 gene. IL-6 is a complex ∼25-kDa cytokine, acting in both pro-and antiinflammatory capacities (Weidhase et al., 2019). Importantly, IL-6 trans-signaling via IL-6/soluble IL-6 receptor (sIL-6R) complexes, but not classic signaling via IL-6/membrane bound IL-6 receptor (IL-6R) complexes, has been shown to prevent the apoptosis of T cells and promote tissue damage (Sommer et al., 2014).
The results of this study suggested that CCP may have an interventional effect on viral infections and lung injury. The pleotropic cytokine, TNF-α, plays a significant role in the pathogens of chronic inflammatory diseases. The MAPK signaling pathway controls diverse cellular processes in response to a variety of extracellular stimuli (Kiss et al., 2019). Taken together, the results suggest that CCP may play a role in the recovery period of COVID-19 by exerting antiviral, bacteriostatic, anti-inflammatory, and immunomodulatory activity. Chinese medicine has always been known for its theoretical understanding and clinical practice of "preventing diseases before disease onset." COVID-19 has a relatively long disease duration, and the recovery period is characterized by a state of low immunity. This study revealed that CCP could regulate immune function through multiple pathways and multiple targets.
We then performed docking studies for TOP, MMP2, MMP9, IFNG, SELE, PLAU, VEGFA, HMOX1, F2, TNF, TP53, PPARG, PIK3CG, IL6, PTGS2, HSP90AB1, EGFR, using the critical ingredients quercetin, kaempferol, and luteolin as ligands. Molecular docking is used to evaluate whether ligands and proteins may bind thermodynamically. Overall, the scores of the active compounds with the key targets were for the majority positive with scores greater than 3, demonstrating that quercetin, kaempferol, and luteolin exhibited good binding properties with VEGFA, TNF, MMP9, and IL-6, respectively, all of which play an essential role in PF.
The pathological process of PF could be roughly divided into three stages. The first stage involves the diffuse damage of vascular endothelial cells and alveolar epithelial cells by pathogenic factors, which initiates the inflammatory immune response. Second, inflammatory cells release a variety of cytokines and inflammatory mediators, expanding tissue damage and causing interstitial hyperplasia. The third step involves the migration and proliferation of fibroblasts and endothelial cells and the metabolic disorders of collagen and other ECM components, which further aggravate inflammatory damage and proliferation in a positive feedback manner. Eventually, the process could lead to the replacement and reconstruction of normal lung tissue. These three processes exist simultaneously, which are interrelated and interact (Fernandez and Eickelberg, 2012;Todd et al., 2012;Kolahian et al., 2016).
VEGF-A is considered a critical factor in the pathogenesis of PF (Medford and Millar, 2006;Barratt et al., 2014). VEGFR is a functional receptor of VEGF and an important target for mediating angiogenesis, which stimulates the proliferation and aggregation of fibroblasts, promotes epithelial-mesenchymal transition, activates multiple abnormal signaling pathways, further induces the formation of fibroblast foci, and causes the activation of the MMP family members to destroy the alveolar structure, promote matrix deposition, and induce scar formation (Leung et al., 1989;Ferrara et al., 2003;Bates, 2010). This process generates other mediators involved in the inflammatory response, such as TNF-α and IL-6, which directly or indirectly promote the synthesis of the ECM through interaction with other cytokines (Le et al., 2014;Hou et al., 2018;Papiris et al., 2018). More importantly, several studies have robustly documented that silencing the expression of TGF-β1 reduces inflammation and slows the progression of PF (Baowen et al., 2010;Liu et al., 2013;Peng et al., 2013), which play a pivotal role in PF (Hu et al., 2018;Li et al., 2018). Findings from our study indicated that CCP could inhibit the expression of VEGF, TNF-α, IL-6, MMP9, and TGF-β1 via the VEGF, Toll-like 4 receptor, MAPK, and TGF-β1 signaling pathways. The potential mechanisms involved in CCP activity are summarized in Figure 8.

CONCLUSION
The findings suggest that CCP treatment of COVID-19 PF associated with SARS-CoV-2 infection involves multiple components and multiple target points as well as multiple pathways. Such proteins should be interesting to future studies that provide a novel direction for the mechanisms of PF associated with SARS-CoV-2 infection development and a new intervention target for clinical investigations, covering these gaps in research to be able to draw more meaningful conclusions about the benefits of CCP. These findings may offer a rationale for further investigations of the anti-fibrotic mechanisms of CCP therapy.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
DJ, XA, and YZ contributed to the drafting of the manuscript. FL and XT revised the manuscript. All authors agreed with the integrity of the study and gave their approval.

FUNDING
This work was funded by the Special Project for Emergency of the Ministry of Science and Technology (2020YFC0845000) and the Traditional Chinese Medicine Special Project for COVID-19 Emergency of National Administration of Traditional Chinese Medicine (2020ZYLCYJ04-1).

ACKNOWLEDGMENTS
We thank the National Administration of Traditional Chinese Medicine for supporting our work.