Analysis of the Molecular Mechanisms of the Effects of Prunella vulgaris against Subacute Thyroiditis Based on Network Pharmacology

Prunella vulgaris (PV) has a long history of application in traditional Chinese and Western medicine as a remedy for the treatment of subacute thyroiditis (SAT). This study applied network pharmacology to elucidate the mechanism of the effects of PV against SAT. Components of the potential therapeutic targets of PV and SAT-related targets were retrieved from databases. To construct a protein-protein interaction (PPI) network, the intersection of SAT-related targets and PV-related targets was input into the STRING platform. Gene ontology (GO) analysis and KEGG pathway enrichment analysis were carried out using the DAVID database. Networks were constructed by Cytoscape for visualization. The results showed that a total of 11 compounds were identified according to the pharmacokinetic parameters of ADME. A total of 126 PV-related targets and 2207 SAT-related targets were collected, and 83 overlapping targets were subsequently obtained. The results of the KEGG pathway and compound-target-pathway (C-T-P) network analysis suggested that the anti-SAT effect of PV mainly occurs through quercetin, luteolin, kaempferol, and beta-sitosterol and is most closely associated with their regulation of inflammation and apoptosis by targeting the PIK3CG, MAPK1, MAPK14, TNF, and PTGS2 proteins and the PI3K-Akt and TNF signaling pathways. The study demonstrated that quercetin, luteolin, kaempferol, and beta-sitosterol in PV may play a major role in the treatment of SAT, which was associated with the regulation of inflammation and apoptosis, by targeting the PI3K-Akt and TNF signaling pathways.


Introduction
Subacute thyroiditis (SAT), which is also called subacute granulomatous thyroiditis, de Quervain's thyroiditis, or giantcell thyroiditis, is the most common cause of thyroid pain. SAT is a self-limited inflammatory thyroid disease possibly related to viral infection that usually presents as a prodrome of lowgrade fever, fatigue, goitre, and pharyngitis symptoms [1].
ere is no definitive cure for painful SAT, but there are effective treatments that will relieve the symptoms and allow the disease to run its course in an asymptomatic fashion, including nonsteroidal anti-inflammatory agents (NSAIDs) and glucocorticoids. Unfortunately, approximately 5% to 15% of patients develop permanent hypothyroidism after recovering full thyroid function within 12 months [2,3] and 1% to 4% relapse after a year. [4].
Prunella vulgaris (PV) has the functions of clearing fire and dispersing knots and swelling, and it has been confirmed that PV suppresses inflammation via several signal transduction pathways [5]. In recent years, PV and its preparations (such as PV oral liquid) have played an important role in the treatment of SAT. PV is recommended in the treatment of SAT for qi stagnation and phlegm-blocking type and the syndrome of deficiency of both qi and yin according to Chinese Medicine Diagnosis and Treatment Scheme of Gall Pain (SAT) by the Chinese State Administration of Traditional Chinese Medicine in 2017 (trial version). According to the expert consensus statement on the treatment of goitre/nodular thyroid disease with PV in clinical practice according to the Chinese Experts of Clinical Application of PV, the use of PV alone to treat SAT with one course each month for 2-3 courses is recommended (recommended classification: B). PV combined with chemicalbased medicine for 4 weeks as a course for 1-2 courses is recommended (recommended classification: B) [6]. Initial treatment combined with PV reduced prednisolone consumption for patients with SAT [7]. However, the molecular mechanisms of the effects of PV against SAT are still unclear.
Network pharmacology, which elucidates the synergistic effects and the underlying mechanism of multiple components and multitargets, has been proven to be a powerful tool for the exploration of Traditional Chinese Medicine [8]. In recent years, network pharmacology has been widely used to investigate the interactions of active ingredients, relevant targets, and molecular mechanisms in TCM [9]. erefore, the present study was based on network pharmacology to elucidate the mechanism of the effects of PV against SAT.
In the present study, network pharmacology analysis, including the identification of active ingredients, prediction of ingredient-related targets and SAT-related targets, construction of a protein-protein interaction (PPI) network, and gene ontology (GO) and KEGG pathway analysis, was subsequently performed to clarify the biological processes related to the target protein involved in the PV-mediated treatment of SAT. e workflow is as follows (Figure 1).

Identification of Components and Targets of PV.
e PVrelated targets were retrieved from the Pharmacology Database and Analysis Platform (TCMSP, http://lsp.nwsuaf. edu.cn/tcmsp.php) based on the term "Xiaokucao" to search for the effective components of PV. en, according to the pharmacokinetic parameters of ADME, components meeting the criteria of both an OB (Oral Bioavailability) ≥30% and a DL (Drug-Likeness) ≥0.18 were defined as active ingredients [10]. OB represents the pharmacological percentage of an oral drug entering the systemic circulation, and a high OB is usually defined as the key index to determine the bioactive molecules in therapeutic drugs. DL is a qualitative concept used to estimate the drug properties of compounds, which is helpful to optimize the pharmacokinetics and drug properties, such as solubility and chemical stability [11]. Subsequently, candidate compounds were input into TCMSP to search for related protein targets.

Prediction of Disease-Related
Genes. Screening of SATrelated genes from the GeneCards database (https://www. genecards.org/) and OMIM database (http://www.omim. org/) was performed. e keywords, "subacute thyroiditis," "subacute granulomatous thyroiditis," "de Quervain's thyroiditis," and "giant-cell thyroiditis," were used, and the targets were the human genes identified in this study. e intersection of compound targets and disease targets is considered to be therapeutic targets.

Network Construction.
Compound-target (C-T), compound-overlapping target (C-OT), and compoundtarget-pathway (C-T-P) networks were constructed by Cytoscape 3.7.2. e topological parameters of the nodes were calculated by a "network analyzer," including the degree centrality (DC) and betweenness centrality (BC). Degree refers to the number of adjacent proteins. BC is another standard measure of node centrality in a network, which measures the total number of nonredundant shortest paths going through a certain node or edge; both parameters are considered to be great predictors of essentiality in the interaction network [12,13]. In the network, nodes represent compounds, targets, and pathways, and edges represent the relationship between compounds and targets and between targets and pathways.
To clarify the direct or indirect regulatory relationship between therapeutic targets, we used the STRING platform (https://string-db.org/) to construct a PPI network, with the species limited to "Homo sapiens" and high confidence (minimum required interaction score ≥0.7). en, the network data were downloaded and input into Cytoscape 3.7.2, and the topological parameters of the nodes were calculated by a "network analyzer." e nodes with DC scores and BC scores greater than 2 times the median were considered as important nodes in the network, namely, as hub targets [14]. In the network, nodes represent targets, and edges represent target interactions.

Pathway and Functional Enrichment Analysis.
To clarify the role of therapeutic target proteins in gene function and signaling pathways, the therapeutic target proteins were submitted to the DAVID database for GO function enrichment analysis and KEGG pathway enrichment analysis. P < 0.05 was considered significant.

Components and Targets of P.
A total of 60 components and 335 targets were retrieved from the TCMSP database. Finally, 11 components were selected as active ingredients based on the criteria of OB ≥ 30% and DL ≥ 0.18 (Table 1), and 126 targets remained after removing duplicate targets, as shown in the C-T network ( Figure 2, Table S1). In addition, 3079 SAT-related targets were selected from the database, and 2207 targets remained after deleting duplicate targets (Table S2). After determination of the intersection of the drug target and disease target, 83 targets were considered to be potential therapeutic targets for treatment of SAT for constructing the C-OT network, which included 94 nodes (11 compounds and 83 targets) and 231 edges (Figures 3 and 4, Table 1). From the perspective of active ingredients, each ingredient was connected with multiple targets. Quercetin (degree � 68) interacted with 68 targets and exhibited the highest correlation with disease targets, and the rest of the active ingredients included kaempferol (degree � 37), luteolin (degree � 32), beta-sitosterol (degree � 27), and stigmasterol (degree � 24). From the perspective of targets, most targets were connected with more than 2 ingredients, and the top 5 targets based on degree were AR (degree � 10), ESR1 (degree � 10), PTGS2 (degree � 8), CA2 (degree � 8), and NOS2 (degree � 7). Moreover, to exhibit biological properties, components and metabolites need to be available in the target tissue. erefore, the biological properties of components also depend on their absorption in the gut and the OB. e top 3

PPI Network and Hub Target Screening.
To obtain information on the predicted interaction, we uploaded the 83 therapeutic targets to the STRING platform, and the results were imported into Cytoscape 3.7.2 to visualize and analyze the interaction network ( Figure 5). ere were 77 nodes and 383 edges in the network. According to the topological parameters of the network, the median DC was 7, and the median BC was 0.00287026. Finally, 18 targets were selected as hub targets based on DC and BC values that were 2-fold higher than the median, accounting for 21.69% of the total targets. As shown in Table 2, TP53, which showed the highest degree, can act on 33 targets, and the rest of the genes ranked according to the degree value were IL6, JUN, MAPK1, MAPK8, TNF, VEGFA, EGF, IL-1β, PTGS2, APP, EGFR, ESR1, CCL2, MAPK14, AR, HMOX1, and NOS3.

GO Analysis of erapeutic Targets.
To elaborate the biological functions of therapeutic targets, we submitted 83 therapeutic targets to the DAVID platform. As a result, a total of 168 GO terms were selected, including 107  Evidence-Based Complementary and Alternative Medicine 5 biological process (BP) terms (Table S3), 36 molecular function (MF) terms (Table S4) and 25 cellular component (CC) terms (Table S5). We screened the top 10 significantly enriched terms in the BP, MF, and CC categories, as shown in Figure 6. e results indicated that therapeutic targets were related to the lipopolysaccharide-mediated signaling pathway, Ras protein, response to hypoxia, reactive oxygen species metabolic process in BP, steroid binding, serinetype endopeptidase activity, core promoter sequencespecific DNA binding, sequence-specific DNA binding, heme binding, cyclin-dependent protein serine/threonine kinase activity in MF, and the extracellular space, cell surface, cytosol, dendritic shaft, and membrane raft in CC.

KEGG Pathway Analysis and C-T-P Network
Construction. A total of 90 KEGG pathways were retrieved (Table S6), and we screened the top 20 pathways after removing uncorrelated pathways such as "bladder cancer" and "Chagas disease" according to a P value < 0.05 (Figure 7). e result indicated that the targets were most closely related to the TNF signaling pathway, and the other pathways ranked by the P value were the HIF-1 signaling pathway, T-cell receptor signaling pathway, PI3K-Akt signaling pathway, sphingolipid signaling pathway, NOD-like receptor signaling pathway, VEGF signaling pathway, and toll-like receptor signaling pathway, which were mainly related to inflammation, hypoxia, immunity, and angiogenesis. A C-T-P network was constructed based on the top 20 pathways and involved targets and the corresponding compounds by Cytoscape 3.7.2 to further illustrate the molecular biological process of PV treatment of SAT ( Figure 8). A total of 65 nodes (11 compounds, 34 targets, and 20 pathways) and 239 edges were obtained. e

Discussion
In the present study, we studied the molecular mechanism of the effects of PV on the treatment of SAT with network pharmacology. A total of 11 components were eventually    [15]. Additionally, it can alleviate the abnormal haemostasis caused by methimazole-induced hypothyroidism in rats by modulating the hydrolysis of adenine nucleotides and nucleosides and consequently reducing platelet aggregation [16]. Kaempferol acted as a competitive TPO inhibitor regulating the synthesis of thyroid hormone in a previous study. Luteolin has been shown to possess anti-inflammatory activity both in vitro and in vivo [17,18], and it has potent anti-inflammatory effects on murine experimental autoimmune thyroiditis [19]. In general, each ingredient is connected with multiple targets, and different components derived from different herbs act on common targets. ese components of phytomedicines exert therapeutic effects through the synergistic action of several chemical compounds acting at multiple target sites. rough the topological analysis of the PPI network, we identified 18 hub genes, including TP53, IL6, JUN, MAPK1, MAPK8, TNF, VEGFA, EGF, IL-1β, PTGS2, APP, EGFR, ESR1, CCL2, MAPK14, AR, HMOX1, and NOS, which can be regarded as the hub targets of PV in the treatment of SAT. ese genes are associated with inflammation, hypoxia, Figure 8: e C-T-P network constructed by Cytoscape. e yellow nodes represent active compounds, the blue nodes represent targets, and red nodes represent pathways. Nodes size and color depth are proportional to their degree. C-T-P network: compound-target-pathway network. apoptotic processes, immune responses, oxidative stress, angiogenesis, and other processes. For example, IL6, together with tumor necrosis factor-alpha (TNF-α), and IL-1β are proinflammatory cytokines that serve as biomarkers for monitoring disease activity and predicting disease severity [20,21]. Jun family members (c-jun, JunB, and JunD) and one Fos family member (c-fos, FosL1, FosL2, and FosB) participate in the formation of AP-1 complexes. c-jun and c-Fos are target genes for the treatment of inflammation, cancer, and vascular reconstruction and can regulate the expression of many downstream genes, such as CCL2 and IL-1β. JunB-independent AP-1 family members mediate the promotion of endothelial proliferation by VEGF, whereas induction of JunB expression primarily mediates VEGF-induced endothelial migration [22]. VEGFA is the most important VEGF, and VEGF is a tyrosine kinase receptor and a key factor for vascular development and formation of new blood vessels (angiogenesis) [23]. EGFR is a 170 kDa transmembrane tyrosine kinase protein, and EGF is one of the ligands of EGFR; EGF can inhibit the uptake and activation of iodine in vivo and in vitro [24], reduce the mRNA and protein expression of TG and TPO, and prevent the release of thyroid hormone to the circulatory system [25]. EGFR triggers downstream signaling pathways such MAPK and PI3K-Akt. rough the GO enrichment analysis of the therapeutic targets, one interesting phenomenon was observed: the therapeutic effect is mainly observed on the cell surface and in the extracellular space, and it is observed to a lesser degree in the cytosol. is result can be explained by the pathogenesis of SAT. is disease is thought to be caused by a viral infection or a postviral inflammatory process, and doublestranded RNA (dsRNA) formed intracellularly in thyrocytes may be a cause of thyroid dysfunction; therefore, autoimmunity may play a secondary role in pathogenesis [26,27]. dsRNAs are recognized as pathogen-associated molecular patterns (PAMPs) by TLRs, and the recognition of viral dsRNA induces type I interferon production and activates innate immune responses. [27,28]. Upon stimulation with TLR by PAMPs, proinflammatory cytokines such as IL-1, TNF-α, and IL-6, which modulate thyroid hormonogenesis, were produced [27,29,30]. In the pathogenesis process, viral recognition and transmembrane signal transduction occur upstream of the whole signal transduction pathway, which play a decisive role in subsequent events. Components of PV may play an important role at this stage.
By integrating the KEGG pathway analysis with the C-T-P network, we speculate that PV exerts its therapeutic effects against SAT mainly through quercetin, luteolin, kaempferol, and beta-sitosterol and is most closely associated with their regulation of inflammation and apoptosis by targeting the PIK3CG, MAPK1, MAPK14, TNF, and PTGS2 proteins and the PI3K-Akt and TNF signaling pathways, according to the topological parameters. e PI3K/Akt pathway functions in the transmission of cell signaling through transduction systems to the cell nucleus, where it plays an important role in cell growth, proliferation, apoptosis, migration, and differentiation [31]. Activation of PI3K/Akt signaling is believed to be a strategy to increase cell survival and proliferation by viruses to suppress apoptosis [32]. Downregulating the phosphorylation level of PI3K/Akt signaling proteins can promote thyroid epithelial cell (Nthy-ori 3-1) apoptosis [33]. A previous study showed that suppressing the PI3K/Akt/mTOR signaling pathway inhibited autophagy in a model of thyroiditis [34]. TNF-α is a multifunctional cytokine with proinflammatory and anti-inflammatory characteristics that plays a critical role in almost every type of inflammatory pathogenesis [35]. It has been demonstrated that flavonoids, including quercetin, kaempferol, and luteolin, possess a benefit in reducing expression of TNF-α and other cytokines. Both quercetin and luteolin at 25 μM effectively decreased TNF-α, IL-6, IFN-c, and IL-1β production in human whole blood incubated with LPS [36]. Moreover, quercetin and luteolin have an effect on the modulation of transcriptional factors; for example, NF-κB and AP-1 are important transcriptional factors involved in the modulation of proinflammatory mediators. Quercetin (100 μM) significantly decreased NF-κB and AP-1 activity in high glucose-induced rat aortic endothelial cells. Luteolin (1-10 μM) significantly inhibited TNF-α and IL-6 production and MAPK (JNK and p38) and transcriptional factor (NF-κB and AP-1) activation in SW982 cells [37]. ese results prove the accuracy of our network pharmacology analysis.

Conclusions
In conclusion, a total of 11 components and 83 overlapping targets between subacute SAT-related and PV-related targets, as well as 18 hub genes and 90 pathways, were retrieved by network analysis. e results of the KEGG pathway and C-T-P network analysis suggested that the active components of PV, including quercetin, luteolin, kaempferol, and beta-sitosterol, may play a critical role in the treatment of SAT by targeting the PIK3CG, MAPK1, MAPK14, TNF, and PTGS2 proteins and the PI3K-Akt and TNF signaling pathways. e present study provides evidence to support the further study of PV for the treatment of SAT. Data Availability e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare no conflicts of interest.

Authors' Contributions
e design of the study was performed by Xia Zhong and Xin Shen, and the study was performed by Xin Shen. Technical assistance was provided by Rui Yang and Jianpeng An. All authors read and approved the final manuscript.
Acknowledgments is work was supported by Guiyang Xintian Pharmaceutical Co., Ltd., Guizhou, China. e authors thank Hongan He, PhD, at Shandong University of Traditional Chinese Medicine for her support in network pharmacology technology. Table S1: 126 targets of all ingredients. Table S2: 2207 targets of SAT. Table S3: 107 biological process terms associated with therapeutic targets.