Rules of Chinese Herbal Intervention of Radiation Pneumonia Based on Network Pharmacology and Data Mining

Objective To explore the mechanism and principles of traditional Chinese medicine (TCM) in the management of radiation pneumonia. Methods The targets of radiation pneumonia were obtained by screening the GeneCards, OMIM, TTD, DrugBank, and HERB databases, analyzing ADME parameters. In addition, compounds and Chinese herbs that can act on the targets were screened from the TCMSP database. The core target compounds for TCM were used to construct the target-compound, compound-traditional Chinese medicine, and target-compound-traditional Chinese medicine networks. These networks were further used to select the core targets, compounds, and TCM. The binding strength between the core targets and compounds was determined using AutoDock Vina. The trajectory for the molecular dynamics simulation was completed by Desmond version 2020. Results A total of 55 active targets in radiation pneumonia were identified. Subsequently, 137 candidate compounds and 469 Chinese herbs were matched. Frequency statistics showed that the Chinese herbs that could interfere with radiation pneumonia were mainly bitter, spicy, and sweet, with both cold and warm properties. Moreover, they mainly belonged to liver and lung channels. The core targets included TNF, IL-6, TGF-β1, and TP53. The most important components were quercetin, resveratrol, and (-)-epigallocatechin-3-gallate. Moreover, the most significant traditional Chinese herbs were Perilla pueraria, ephedra, Lonicerae japonicae, and sea buckthorn. Furthermore, analysis of 222 sets of receptor-ligand docking results suggested that the compounds had good docking activity to their core targets. By combining the docking binding energy, we determined that the chemical compounds had strong binding energy to the targets. Conclusion Using network pharmacology, we explored the potential mechanism of TCM in the treatment of radiation pneumonia. The general rules for application of TCM in the treatment of radiation pneumonia were summarized. This study provides baseline information for future research on the development of TCM for the management of radiation pneumonia.


Background
Radiation pneumonitis (RP), an inflammatory reaction in normal lung tissues, is caused by injuries in the radiation field after radiotherapy for thoracic or other malignant tumors. Clinically, RP is the first stage of radiation-induced lung injury (RILI), a common radiotherapeutic complication in tumor patients. Acute radiation pneumonitis (ARP) occurs within 1-3 months after radiotherapy. Typically, it is characterized by acute onset, high fever, choking cough with About 50% of tumor patients in the United States choose and receive radiotherapy, with 70-90% of them being thoracic cancer patients [5]. Nuclear radiation accidents and pretreatment with bone marrow transplants have also been associated with RP occurrence. us, RP is a major clinical concern.
Conventional treatment options include antibiotics along with adrenocorticosteroids. Patients with severe dyspnea need continuous oxygen therapy or assisted ventilation. ese therapeutic measures can temporarily relieve disease symptoms; however, their long-term efficacy is not sufficient. Moreover, the adverse effects of hormones, such as hormone-mediated immune inhibition, induction of double infections, and osteoporosis, limit their clinical applications.
Various drugs with reactive oxygen scavenging effects, such as sulfhydryl compounds, antioxidant enzymes and their analogues, and plant antioxidants [6,7], can be used as radioprotective agents to remove free radicals produced by radiation or other cytotoxic drugs in normal tissues. However, these treatment forms are costly and may cause serious toxic side effects. e pathogenesis of RP has been well elucidated, but, currently, there are no effective preventive and treatment options for the disease. erefore, it is vital to find targets or drugs with protective effects on normal lung tissues to reduce radiotherapy-associated lung injuries in lung cancer patients [8]. Traditional Chinese medicine (TCM) has great significance in preventing and improving the clinical symptoms and severity of RP. Furthermore, TCM can act on multiple targets and offer a comprehensive advantage for diseases with complex pathogenesis.
In this study, through the network pharmacology approach, the RP-related targets were taken as entry points and matched to their corresponding ligand compounds in Chinese medicine. We aimed at investigating the complex connection between targets, compounds, and Chinese medicine. Moreover, we assessed the mechanisms of action and general rules of Chinese medicine in RP. Our findings provide a theoretical basis for subsequent exploration of Chinese medicine, therapeutic strategies, and clinical integration of Chinese medicine into Western medicine in RP management. A flow chart of this study is shown in Figure 1.

Selection of Alternative Compounds and "Compound-Target" Network Construction.
e Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) (https://tcmspw.com/) [15] was used to obtain compounds that could act on RP-related targets.
Candidate compounds were selected based on the ADME index and oral bioavailability (OB) ≥30% and drug half-life (HL) >4 as well as drug likeness (DL) ≥0.18 [16]. After initial screening, deleted components were checked one by one after which relevant active ingredients were added. Given that composition information in the TCMSP database has computer prediction results, some of the data may not correspond to actual applications. us, screened compounds were used as candidate compounds, which were then imported into Cytoscape 3.7.2 software [17] along with targets to construct a "target-compound" network. ereafter, topological properties of the network nodes were analyzed.

Herbal Acquisition and "Target-Compound-Chinese
Medicine" Network Construction.
e TCMSP database was used to acquire TCMs containing alternative compounds and to construct the alternative "compound-Chinese medicine" network. In conjunction with "compound-target" network, topological parameters for each node in the network were determined using the NetworkAnalyzer module of Cytoscape 3.7.2 to identify key nodes. en, core drugs were selected to construct the "target-compound-Chinese medicine" network in order to comprehensively evaluate the efficacy of different compounds and Chinese medicine in RP treatment.

Molecular Docking Assessment.
High ranking target proteins were screened by Cytoscape "target compound network". en, 3D crystal structures of target proteins were downloaded from the RCSB database (https://www.rcsb.org/ pdb) and converted into PDB files for storage using Open Babel 3.1.1 (https://openbabel.org/wiki/Main_Page) [18]. en, AutoDock Tools 1.5.6 (https://autodock.scripps.edu/ resources/tools) were adopted for dehydrating and hydrogenation, calculating the Gasteiger charge, and storing it as a PDBQT file [19]. e potential core distributer was minimized by energy processing and the ligand atom type assigned to calculate the charge, which was stored in a PDBQT format. Molecular docking and docking scoring were performed using Moe2019 [20]. Active ingredients and targets with good binding activities were screened by Affinity. Docking score ≤−4.25 kcal/mol implies binding activities between the ligand and target and score ＜−5.0 kcal/ mol indicates a good binding activity, while score ≤−7.0 kcal/ mol implies strong docking activities [21].

Molecular Dynamics
Simulation. MD simulations of docked complexes were performed using Desmond version 2020.
e OPLS3e force field was used to initiate MD simulation, after which the system was solvated using a TIP3 water model. On postpreparation of the system, the production run was performed for 200 ns with a time step of 1.2 fs. Trajectory recording was done for every 200 ps, summing up to a recording of 10,00 frames. e root mean square deviation (RMSD) was calculated for backbone atoms and graphically analyzed to understand protein-ligand interactions. Root Mean Square Fluctuation (RMSF) for every residue was calculated to understand the major conformational changes in residues between the initial and dynamic state.

Results of Target Acquisition.
In this study, 96, 7, 2, and 0 RP-related targets were, respectively, obtained from the GeneCards, OMIM, TTD, and DrugBank databases. e targets were complemented with the Herb database, and duplicate values were removed after which the targets were matched with the TCMSP database. Unmatched targets were eliminated. Finally, 55 related targets were obtained and standardized in the UniProt database.

Candidate Compounds and the Target-Compound
Network. Fifty-five potential targets were matched with 12,144 small molecule compounds from the TCMSP database and a total of 137 potential compounds obtained. Figure 2 shows the target-compound network constructed using potential targets and candidate compounds. e Step 1 : Disease Targets Collection and Standardization Step 2 :

Connection of Relative Compounds and Herbs
Acting on Targets Step 2 : Analysis of Potential Core Drugs and Compounds

Targets-Compounds-Herbs
Network Core Drugs and Compounds Screening, molecular docking and dynamics simulation

Candidate Compounds and Target-Compound Network.
Based on the 137 candidate compounds, screening of TCMSP and chemical databases revealed 469 Chinese herbs. e compound-Chinese medicine network was constructed based on the relationship between compounds and traditional Chinese medicine. e network had 605 nodes and 2625 edges. Median degree values in the network were calculated twice and a value of 20 is obtained. A degree value >50 indicates a potential core compound. Sixteen potential core compounds were identified in the network. Table 1 shows the top 5 potential core compounds (beta-sitosterol, palmitic acid, quercetin, kaempferol, and luteolin) ranked according to their degree values. Degree, betweenness centrality, and closeness centrality of potential core compounds were calculated using Network Analyzer in Cytoscape 3.7.2. e degree, betweenness centrality, and closeness centrality of beta-sitosterol were 247, 0.269, and 0.493, respectively. Beta-sitosterol was predicted to be the main compound that can be used to intervene RP because its topological parameters had the highest value. Palmitic acid ranked second with a degree, betweenness centrality, and closeness centrality of 239, 0.274, and 0.487, respectively. Quercetin ranked third with a degree, betweenness centrality, and closeness centrality of 289, 0.139, and 0.445, respectively. e top 10 types of traditional Chinese medicines ranked according to their degree values were Yin-XingYe, YuanSui, GeHua, MaHuang, ZiSu, SangYe, HongHua, ShaJi, LianQiao, and JinYinHua. ereafter, based on bridging level between compounds, traditional Chinese medicine targets that were related to RP treatment were identified. e top-ranking compounds were Zi Su, Ge Hua, Ma Huang, and Jin Yin Hua (Table 2). us, RP treatment might involve regulation of different targets and pathways by traditional Chinese medicine.
To understand the relationship between the targets, compounds, and TCM applied in the treatment of RP, a target-compound-traditional Chinese medicine network was generated ( Figure 3). e degree values of objects included in the network were higher than 5. In Figure 3, the diamond, hexagon, and V shapes represent traditional Chinese medicine, compounds, and targets, respectively. In addition, the icon size was positively correlated with the degree values. Calculation of the median degree value of the compounds was performed thrice and a value of 41.5 was obtained. Core compounds were those with a degree value equal to or greater than 41.5. As a result, 37 compounds were identified as core compounds with the top 5 being betasitosterol, palmitic acid, quercetin, kaempferol, and luteolin.
Out of the 222 identified groups, 95 were present in the target-compound network and were therefore referred to as the in-network combination. Notably, TNF-Rutin had the highest molecular docking score (−9.34 kcal/mol), followed by TP53-EGCG (−8.31 kcal/mol), and then TNF-EGCG (−8.29 kcal/mol). On the other hand, TNF-Alpha-humulene had the lowest molecular docking score (−4.26 kcal/mol). e average affinity score of the above combinations was −6.34 kcal/mol, indicating high interaction between compounds and targets in the target-compound network. Additionally, 127 new combinations were discovered outside the network, among which TP53-Rutin had the highest affinity score (−10.26 kcal/mol), followed by CDKN1A-Rutin (−8.76 kcal/mol) and TP53-rosmarinic acid (−8.39 kcal/mol). Further, VEGFA-3,4,5-trihydroxybenzoic acid had the lowest score (−4.3 kcal/mol) with a mean affinity score of −6.05 kcal/mol. ese results demonstrated active binding between the core compounds and core targets. Figure 4 shows that Rutin, rosmarinic acid, (-)-epigallocatechin-3-gallate, and puerarin showed strong binding to other core targets.
In addition, nine ideal combinations were identified in the molecular docking experiments based on the affinity scores and the degree values of the target-compound-traditional Chinese medicine network. e three-dimensional and two-dimensional molecular docking models are shown in Figure 5. e models show that the ligands were embedded in the active pockets of the targets. In addition, the ligands interacted with multiple residues on the targets through hydrophobic interactions and hydrogen bonds.

Molecular Dynamics Simulation.
e protein alpha root mean square deviation (RMSD) data of Rutin and IL-6 and epigallocatechin and TNF as well as Rutin and TNF are presented in terms of proteins and ligand as shown in ese findings show that a stable conformation was achieved during MD simulation.

Statistics and Frequency Analysis of Flavour, Nature, and
Meridian. Information on flavour, nature, and meridian was collected from 469 traditional Chinese herbs. e results of frequency statistical analysis conducted using Excel program showed that, in terms of nature, bitter (225), acrid (183), and sweet (174) had the highest ranking. On the other hand, the proportion of warm (25.6%) and cold (25.2%) herbs was the highest. In addition, the highest ranked meridians were liver (17.1%), stomach (13.6%), and spleen (11.9%), which were involved in the treatment of RP. e results of frequency statistical analysis are shown in Figure 7.

Discussion
Radiation pneumonitis is a radiotherapy-associated complication during malignant tumor treatment. However, RP is not found in the well-known TCM literature, even though according to its symptoms, it belongs to the "cough," "lung carbuncle," "atrophic lung disease," and "Pulmonary palsy" categories. Clinically, radiation is believed to be the main source of heat toxicity. Lungs are too delicate for excess cold and heat. During radiotherapy, heat toxicity accumulates in the lungs, leading to considerable organ damage. A combination of deficiency-heat and blood stasis can further destroy the lungs, resulting in the loss of normal lung functions. e tumor itself can destroy the body and heattoxicity invades body, leading to qi collapse, which results in atrophic lung disease. erefore, heat toxicity, deficiency of Yin, and static blood are the main pathological features of RP. Lung disease pathogenesis involves blood stasis as well as the deficiency of both qi and yi. e main treatment modes involve heat clearance, toxin removal, nourishing qi and yin,  activating blood, and resolving stasis [22]. However, often, different doctors have distinct views. For instance, Professor Wang suggests that although radiation pneumonia has such pathological features as heat toxicity, lung dryness, yin and qi deficiency as well as fluid deficiency, it is caused by a wind pathogen, it occurs in the lungs, the primary cause is liver, insufficient lung fails to restrict liver are considered to be one of the pathogenesises. erefore, the main treatment forms involve dispersing wind and relieving cough [23]. Besides, Professor Shu found that RP is more likely to occur during seasonal changes or in the influenza period and is usually caused by respiratory tract infections. erefore, Professor Shu cleared the heat and removed toxins when treating patients who had received chest radiotherapy. In addition, he dispersed the wind and cooled the blood to prevent the disease [24]. We found that bitter, sweet, and spicy traditional Chinese herbs can potentially treat RP. A combination of cold and warm medicines may be more effective. e main channels belong to the liver meridian and lung meridian.
ese results correspond to clinical cognitive of lung injury caused by radiation therapy. Medicines that are warm or cold and spicy, sweet, or bitter have been reported to be more effective in clearing heat and removing toxins, nourishing qi and Yin, activating blood and resolving stasis, clearing heat, and promoting fluid production [25]. e bitter taste can resolve dampness and preserve yin and is often used to manage heat syndrome in TCM. Furthermore, acrid can be divergent, promoting the flow of qi and blood circulation. Additionally, it can diverge Yu fire and can also work with the bitter cold medicine to prevent fire. Sweet drugs can be tonic, slow, urgent and moderate. After radiotherapy, the human body consumes more energy, which manifests as deficiency of both qi and yin. Sweet and cool drugs can nourish qi and yin. Based on the four properties, including cool, cold, warm, and hot, most of the traditional Chinese medicines are warm and cold. Cold herbs clear heat, reduce fire, cool the blood, and remove toxins. A combination of bitter and cold herbal medicines clears heat and resolves dampness while a combination of sweet and cold medicines nourishes yin and clears heat, consistent with pathogenesis of RP lung heat and yin injury. According to the meridian tropism theory, RP pathogenesis involves wood depression transforming into wind and wood fire tormenting metal. Wind medicine can enter the liver meridian and help to grow upwards. Notably, leaves of ephedra and loquat can regulate dispersion in the lungs.

Evidence-Based Complementary and Alternative Medicine
Modern medicine has not conclusively determined the mechanisms of RP, and a number of theories have been put forward. For instance, the cytokine theory suggests that RP occurs due to radiation-associated activations of macrophages, mast cells, lung interstitial cells, alveolar epithelial cells, and other target cells in the irradiation field. e activated cells release large amounts of proinflammatory cytokines that locally accumulate, thereby regulating and accelerating inflammatory responses. Cytokines that are closely related to RP occurrence and development include Transforming Growth Factor-β1 (TGF-β1), Tumor Necrosis Factor-α (TNF-α), Interleukin-6 (IL-6), and Platelet-Derived Growth Factor (PDGF). Notably, TGF-β1 is an important cytokine for predicting RP occurrence. Radiation causes the TGF-β1 produced during inflammatory responses in lung tissues to activate the phosphatidylinositol 3-kinaseserine/threonine protein kinase (PI3K/AKT) signaling pathway, which promotes alveolar epithelial cell transformations into lung interstitial cells [26], further causing RILI. Elevated TGF-β levels can also be used to diagnose earlystage RP and late-stage radioactive pulmonary fibrosis. Moreover, IL-6 has dual effects of inducing and inhibiting inflammation. Zhang et al. reported that RP patients have elevated IL-6 levels, relative to those without the disease, suggesting that IL-6 is involved in RILI processes after radiotherapy [27]. A previous study reported a significant increase in IL-1β levels after radiation [28]. Biologically, TNF comprise a family of cytokines that can damage tumor cells and cause necrosis. Among the TNFs, TNF-α is an initiator of the cytokine regulatory network in the body [29]. TNF-α also plays an important role in inducing apoptosis, regulating immune responses, and initiating as well as maintaining the occurrence and development of RP. Rübe et al. [30] showed that the timepoint at which plasma TNF-α levels in lung cancer patients increased before and after radiotherapy coincided with the period during which RP occurred, indicating that, after irradiation, TNF-α is involved in inflammatory responses in the lungs.
In the target-compound network constructed in this study, the highest ranked targets included TP53, VEGFA, SOD, and CAT, in addition to TGF-β1, IL-6, and TNF. It was also predicted that these targets are important in regulating RP. TP53 is a tumor suppressor gene that is expressed in low levels in normal cells, but highly expressed in malignant tumors. VEGFA, a member of the Platelet-Derived Growth Factor (PDGF)/Vascular Endothelial Growth Factor (VEGF) family, is the main inducer of angiogenesis. In addition, PDGF promotes pulmonary vascular smooth muscle cell proliferation and angiogenesis. Radiotherapyinduced liver fibrosis is often accompanied by angiogenesis [31], suggesting that VEGFA may play a significant role in RP. Oxidative stress is one of the mechanisms involved in RILI pathogenesis [32]. erefore, it is possible that radiation-induced cytotoxicity is mediated by Reactive Oxygen Species (ROS). Notably, SOD and CAT constitute the body's first line of defense against reactive oxygen species.   TP53  TNF  CDKN1A  TGF-β1  IL6  VEGFA   alpha-

Evidence-Based Complementary and Alternative Medicine
Additionally, given that its overexpression significantly reduced inflammation in the lungs and oxidative damage in RILI mice models, SOD has some anti-inflammatory effects and can inhibit lung fibrosis [33]. Target-compound screening revealed that quercetin, resveratrol, and epigallocatechin gallate (EGCG) had a higher connectivity degree. Quercetin, resveratrol, and EGCG have potent antioxidant, anti-inflammatory, and antitumor activities [34][35][36]. Quercetin is a polyphenolic flavonoid compound that can significantly reduce inflammatory exudation of radiation-induced lung injury and has a certain protective effect against RP [37]. Resveratrol, a natural polyphenol that occurs as a phytoalexin, has certain protective effects against RP and enhances the radiosensitivity of lung cancer cells [38,39], which may be related to its ability to suppress TGF-β1 levels. EGCG, the major polyphenolic component of dried green tea extracts, can reduce serum levels of inflammatory cytokines, IL-6, IL-10, and TNF-α in lung injury. In radiation-induced lung injury, EGCG suppresses early inflammatory responses and late pulmonary fibrosis [40]. Moreover, EGCG can reduce ionizing radiation-mediated damage by scavenging for free radicals, resisting oxidative stress, and hindering apoptosis [41][42][43]. e compound-Chinese herb network in this study showed that palmitic acid and kaempferol are important compounds that can interfere with RP progression through multiple targets. However, it has not been established whether they can affect RP development. Based on degree value and number of targets, the most important traditional Chinese herbs were Perilla pueraria, ephedra, honeysuckle, sea buckthorn, Ginkgo biloba, and cuspidatum. Pueraria contains flavonoid components such as quercetin and genistein. Previous studies reported a significant increase in SOD and CAT activities in high dose Pueraria polysaccharide group after irradiation of mice with X-rays, implying that the Pueraria polysaccharide has a certain preventive and protective effect against radiationassociated damage [44]. Biologically, TNF-α is a negative regulator of hematopoiesis. Treatment of radiation-injured mice with flavonoids from the sea buckthorn suppressed TNF-α levels [45,46]. TNF-α can also suppress radiationinduced bone marrow cell apoptosis, which provides a basis for applications of traditional Mongolian medicine, Hippophae, in radiation damage. Moreover, the Ginkgo biloba extract increased the number of lymphocytes in irradiated mice, mitigated radiation-associated damage to the liver synthesis function, increased SOD as well as GSH activities, and suppressed malondialdehyde (MDA) levels, a product of membrane lipid peroxidation. Honeysuckle and Polygonum cuspidatum have the functions of clearing heat and removing toxicity [47]. Resveratrol from Polygonum cuspidatum was shown to inhibit radiation-induced apoptosis of spleen cells in mice [47]. Additionally, a high dose of the ephedra compound suppressed VEGF levels in tumor tissues of lung cancer mice models, thereby inhibiting tumor angiogenesis [49]. erefore, ephedra has the potential for RP prevention and treatment. Studies on prevention and treatment of RP using Perilla or on the role of Perilla in radiation resistance are few. However, its targets have many intersections with RP disease targets, suggesting that it is a promising drug candidate for RP prevention and treatment. Overall, most of the important traditional Chinese herbs in the network diagram are used for relieving exterior disorders. erefore, RP treatment should focus on early exiting of evil, guiding it from its entry to its exit, and pay attention to the existence of exterior syndromes of TCM. Moreover, it is possible to prevent and cure RP using ROS-targeted treatments, inhibiting the release of inflammatory factors and activation of related signaling pathways.
Molecular docking showed that Rutin, rosmarinate, and Puerarin had good molecular docking scores. Rutin regulates the immune system by inhibiting radiation-induced oxidative and inflammatory stress, which helps in minimizing radiation-induced pneumonia and fibrosis [48]. Rosmarinic acid suppresses intracellular production of ROS, VEGF, and IL-8 levels [49]. Rosmarinic acid may also play a 16  role in interfering with RP progression. Puerarin exhibited protective effects in experimental rats with ionizing radiation injury [50]. e study also showed that, after exposure to radiation, Puerarin can effectively resolve the decrease in red and white blood cells in peripheral blood of rats.
Puerarin-associated protective effects may be related to elevated SOD activities and suppressed MDA levels. ese findings imply that a large number of active components in Chinese medicines can interfere with RP occurrence and development. However, the associated mechanisms of action have not been elucidated. Molecular dynamics simulation results showed that both compounds could stably bind the receptor, proving the reliability of the predicted efficacy results.
Based on the network pharmacology method, we established an RP-related target-compound-Chinese medicine network. In addition, the potential Chinese herbs that may interfere with the RP process and the possible molecular basis as well as mechanisms of action were explored. e 55 RP targets were matched to 137 candidate compounds and 469 Chinese herbs. e most important Chinese herbs were identified to be Perilla, Pueraria, ephedra, honeysuckle, sea buckthorn, Ginkgo biloba, and cuspidatum, highlighting some new drug candidates for RP treatment. Nonetheless, their mechanism of actions should be explored further. Despite the insightful findings, this study has some limitations. For example, due to incomplete information in the database, some active ingredients of core targets may have been omitted during the screening process. erefore, the scope of database screening should be expanded to explore the role of important target components in improving the pathological processes of RP. is will help in identification of more comprehensive drug candidates and better explain the mechanisms of drug intervention in RP.
Data Availability e datasets used and/or analyzed in the current study are available from the corresponding author upon reasonable request. Disclosure e funding agencies had no role in the study design, data collection, analysis, manuscript preparation, or the decision to publish.