Abstract Background Although coronavirus disease 2019 (COVID-19) pandemic is still rage worldwide, there are still very limited treatments for human coronaviruses (HCoVs) infections. Xiaochahu decoction (XCHD), which is one of the traditional Chinese medicine (TCM) prescriptions in Qingfeipaidu decoction (QFPDD), is widely used for COVID-19 treatment in China and able to relieve the symptoms of fever, fatigue, anorexia, and sore throat. To explore the role and mechanisms of XCHD against HCoVs, we presented an integrated systems pharmacology framework in this study. Methods We constructed a global herb-compound-target (H-C-T) network of XCHD against HCoVs. Multi-level systems pharmacology analyses were conducted to highlight the key XCHD-regulated proteins, and reveal multiple HCoVs relevant biological functions affected by XCHD. We further utilized network-based prediction, drug-likeness analysis, combining with literature investigations to uncover the key ani-HCoV constituents in XCHD, whose effects on anit-HCoV-229E virus were validated using cytopathic effect (CPE) assay. Finally, we proposed potential molecular mechanisms of these compounds against HCoVs via subnetwork analysis. Results Based on the systems pharmacology framework, we identified 161 XCHD-derived compounds interacting with 37 HCoV-associated proteins. An integrated pathway analysis revealed that the mechanism of XCHD against HCoVs is related to TLR signaling pathway, RIG-I-like receptor signaling pathway, cytoplasmic DNA sensing pathway, and IL-6/STAT3 pro-inflammatory signaling pathway. Five compounds from XCHD, including betulinic acid, chrysin, isoliquiritigenin, schisandrin B, and (20R)-Ginsenoside Rh1 exerted inhibitory activity against HCoV-229E virus in Huh7 cells using in vitro CPE assay. Conclusion Our work presented a comprehensive systems pharmacology approach to identify the effective molecules and explore the molecular mechanism of XCHD against HCoVs. Supplementary Information The online version contains supplementary material available at 10.1186/s12906-023-04024-6. Keywords: Xiaochaihu decoction, Human coronavirus, Systems pharmacology, Molecular mechanism, HCoV-229E virus Introduction Coronavirus (CoV) is a kind of single-stranded RNA virus that can infect many animal species which can lead to different degrees of lesions in the respiratory tract, liver, intestines and nervous system [[41]1]. CoV contains a total of 4 subfamilies, including α-CoV, β-CoV, γ-CoV and δ-CoV, of which α and β subfamilies are capable of infecting mammals including humans [[42]2]. At present, the human coronaviruses (HCoVs) include HCoV-OC43 (β-CoV) [[43]3], HCoV-229E (α-CoV) [[44]4], HCoV-NL63 (α-CoV) [[45]5], and HCoV-HKU1 (β-CoV) [[46]6]. In the past 20 years, three highly pathogenic CoVs prevailed worldwide, which are severe acute respiratory syndrome coronavirus (SARS-CoV), Middle East respiratory syndrome coronavirus (MERS-CoV) and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) [[47]7]. As of March 10, 2023, the number of confirmed cases of the coronavirus disease 2019 (COVID-19) has reached up to 651 million with a cumulative death of exceeded 6.8 million globally [[48]8] ([49]https://coronavirus.jhu.edu/map.html). Thus, it is of great urgency to develop the new effective anti-HCoVs drugs for patient therapy. Preliminary clinical practice evidences showed that traditional Chinese medicine (TCM) prescriptions, especially those consisting of a combination of herbs, have achieved beneficial effect for COVID-19 patients by shortening of hospitalization duration and reducing the chance of getting complications as well as the mortality rate [[50]9, [51]10]. Qingfeipaidu decoction (QFPDD) consisted of four classical TCM prescriptions [[52]11], is currently used as a common prescription in the COVID-19 diagnosis and treatment scheme in China. From a retrospective multicenter study on 782 COVID-19 patients from 54 hospitals in nine provinces of China, compared with treatment with QFPDD initiated after 3 weeks of infection, treatment started less than 1 week, 1–2 weeks, or 2–3 weeks had a significantly shorter recovery time, with adjusted hazard ratio of 3.81 (2.65–5.48), 2.63 (1.86–3.73), and 1.92 (1.34–2.75), respectively [[53]12]. As one of the four prescriptions in QFPDD, Xiaochaihu Decoction (XCHD) plays the roles of relieving symptoms of fever, fatigue, anorexia and sore throat after SARS-CoV-2 infection [[54]13]. Clinical observation indicated that modified XCHD exhibited beneficial effects on COVID-19 patients, with a response rate of 96.43% after 1 to 2 weeks of treatment. Early XCHD treatment can completely relieve the condition of patients with mild and moderate symptoms, and prevent them progressing into severe stage [[55]14]. XCHD is composed of seven medicinal herbs, which are Bupleurum chinense DC. (Chaihu), Scutellaria baicalensis Georgi. (Huangqin), Pinellia ternata (Thunb.) Breit.(Banxia), Zingiber officinale Roscoe (Shengjiang), Panax ginseng C.A.Mey. (Renshen), Glycyrrhiza uralensis Fisch. (Zhigancao) and Ziziphus jujuba Mill. (Dazao). Previous pharmacological studies showed the effects of XCHD on resisting various RNA and DNA viruses (including coronaviruses) infection, improving immune function and restoring body temperature by hypothalamus regulating [[56]15, [57]16]. Chaihu has been demonstrated to curb the infections of influenza virus, hepatitis virus, and other viruses [[58]17, [59]18]. Huangqin, was proved to enhance leukocyte phagocytosis, increase free antibodies, inhibit the release of active substances and regulate immune function [[60]19, [61]20]. Although the pharmacological actions of XCHD make it a treatment option for COVID-19, there is a great challenge to evaluate the efficacy, identify the functional constituents contained, and explore its molecular mechanisms against COVID-19. TCM is characterized by complex components, multiple targets and synergistic actions. Systems pharmacology had been recentely developed in TMC studies, it provided a network perspective to explore the relationship between components and targets [[62]21]. It has been shown that systems pharmacology gave an effective approaches to identify the bioactive compounds, predict the corresponding targets and elucidate the molecular mechanisms about how does TCM work on different diseases [[63]22, [64]23]. For instance, by using this strategy, it uncovered the mechanism of Huanglian-Wuzhuyu herb pair in treating nonalcoholic steatohepatitis and predicted active ingredients, it also explained mechanism of Lianhuaqingwen capsule in treating COVID-19 [[65]24, [66]25]. In this study, we presented a systems pharmacology-based framework to identify the effective components of XCHD and explore the underlying mechanism of XCHD against HCoVs (Fig. [67]1). Briefly, we first constructed a global herb-compound-target (H-C-T) network for XCHD, which integrated compounds from XCHD, known targets from published experimental literatures, and computationally putative targets predicted from balanced substructure-drug-target network-based inference (bSDTNBI) method [[68]26, [69]27]. Subsequently, a gene set of HCoVs-associated proteins was mapped into the H-C-T network to determine the HCoVs-associated targets that can be regulated by XCHD. Then we applied specific anti-HCoV compound-target (C-T) network, protein–protein interaction (PPI) network, gene set enrichment, and integrated pathway analysis to comprehensively explore the potential biological functions and signaling pathways affected by XCHD. Furthermore, in silico strategy was utilized to identify the active anti-HCoVs components in XCHD by integrating network-based analysis and drug-likeness prediction. Finally, in vitro HCoV-229E virus-induced cytopathic effect assay was carried out to validate the anti-HCoV activity of predicted compounds, while subnetwork analysis was used to investigate the specific synergistic mechanisms of the main active compound candidates in XCHD. Fig. 1. [70]Fig. 1 [71]Open in a new tab Schematic of the systems pharmacology infrastructure for uncovering the molecular mechanism of XCHD against HCoVs Materials and methods Compounds and virus XCHD is composed of seven medicinal herbs, including Chaihu, Huangqin, Banxia, Shengjiang, Renshen, Zhigancao and Dazao. The scientific species names of each herb are shown in Table [72]1. Table 1. The scientific species names of herbs in XCHD Herb Latin Binomial Name Chaihu Bupleurum chinensis DC Huangqin Scutellaria baicalensis Georgi Zhigancao Glycyrrhiza uralensis Fisch Banxia Pinellia ternata (Thunb.) Breit Renshen Panax ginseng C. A. Meyer Shengjiang Zingiber officinale Roscoe Dazao Ziziphus jujuba Mill [73]Open in a new tab Total 30 compounds from these herbs for in vitro assays were commercially obtained from Topscience Biochemical Technology Co., Ltd. The quality control assessment was performed by NMR or HPLC–MS to assure the purities are all greater than 95%. Huh7 cells and HCoV-229E were kindly provided from CAMS Key Laboratory of Antiviral Drug Research, Institute of Medicinal Biotechnology, Chinese Academy of Medical Sciences and Peking Union Medical College. Collection of herbal compounds in XCHD The chemical structures of all compounds in XCHD were collected from the following databases: Traditional Chinese medicine integrative database (TCMID) [[74]28], TCM Database@Taiwan [[75]29], Traditional Chinese medicine systems pharmacology database and analysis platform (TCMSP) [[76]30], Database of Traditional Chinese Medicine on Immuno-Oncology (TCMIO) [[77]31], and the database and analytical system for network pharmacology analysis for TCM preparations (TCM-MESH) [[78]32]. All compounds were converted to InChIKey and SMILES formats by Open Babel (version 2.3.2) [[79]33]. Compounds with identical structures were merged. The above databases were searched until April 2020. Target identification of compounds in XCHD Both known and predicated targets of XCHD were included in the current study. The known targets were extracted from the previous integrated database [[80]26], which contains 7,030 experimentally validated compound-target interactions (CTIs) collected from ChEMBL (v21) and Binding DB. The predictive network model was established by bSDTNBI method to prioritize potential targets for natural products by resource-diffusion processes of the substructure-drug-target network [[81]26]. The tunable parameters α (initial resource allocation of different node types), β (weighted values of different edge types), γ (influence of hub nodes) and k (number of resource-diffusion processes) were set as 0.1, 0.1, -0.5, and 2, respectively. The substructure items of each compound were calculated using molecular fingerprint Klekota − Roth from PaDEL-Descriptor (version 2.18) [[82]34]. Collection of HCoV-associated genes We comprehensively retrieved the literatures to obtain the human genes associated with multiple types of HCoVs, including HCoV-OC43, HCoV-229E, HCoV-NL63, HCoV-HKU1, SARS-CoV, MERS-CoV and SARS-CoV-2. The names of the collected genes were standardized into gene symbol according to GeneCards [[83]35] (retrieved until Jan 2021) and UniProt [[84]35, [85]36] (retrieved until Jan 2021), while the duplicates were removed. Drug-likeness screening In this study, drug-likeness analysis was conducted using a classification model based on random forest (RF) method, the analysis was available at ADMETlab platform ([86]https://admetmesh.scbdd.com/) and completed in April 2021. Specifically, the RF model was trained by 6,731 positive samples from DrugBank and 6,769 negative samples from ChEMBL with IC[50] or Ki values < 10 μm. The obtained model has good ability to generalize the new chemical entity with classification accuracy of 0.800 and AUC score of 0.867 on external test set. More detailed information can be found in previous study [[87]37]. Network construction In our study, we generated three types of networks to explore the molecular mechanism of XCHD against HCoVs, including compound-target (C-T) networks, target-pathway network and protein–protein interaction (PPI) network. These networks were constructed by Gephi (v0.9.2, [88]https://gephi.org/) and Cytoscape (v3.2.1, [89]http://www.cytoscape.org/). The compounds, pathways and genes (targets) were represented as nodes. The interactions were denoted as edges. The degree of each node was defined by the number of edges linked to it, which represents the hierachy of the node in the network. For the PPI network, the functional relationships among interacting proteins were generated through STRING database [[90]38]. The protein type was defined as "Homo Sapiens", while the reliability score of the PPI edge required interaction score was set as greater than 0.4. Gene enrichment analysis To explore how do the targets of XCHD exert their anti-HCoVs effect through pathway regulation, we annotated the biological functions of the HCoVs targets of XCHD to find out the potential signaling pathways and biological processes. Gene ontology (GO) term and KEGG pathway enrichment analysis ([91]https://www.kegg.jp/kegg/kegg1.html) [[92]39] were performed by The Database for Annotation, Visualization and Integrated Discovery (DAVID 6.8 database, [93]https://david.ncifcrf.gov/) [[94]40] (retrieved until Apr 2021). The biological processes, molecular functions, cellular components and signaling pathways with p < 0.05 were considered as statistically significantly enriched. Cytotoxicity test and cytopathic effect (CPE) reduction assay Huh7 cells seeded in 96-well culture plates were incubated in a 37 °C and humidified 5% CO[2] atmosphere until reaching 80% confluency. For cytotoxicity test, the untreated cells were used as a reference; other cells were treated with the test drugs of 8 serially diluted concentrations. For CPE assay, the cells were either infected with 100 TCID[50] HCoV-229E only or both treated with 100 TCID[50] HCoV-229E and the different drugs. Ribavirin treatment was used as the positive control for HCoV-229E infection inhibition. The cells were incubated until the CPE of viral wells reached 4 + (0 means no CPE; 1 + means CPE is 1%-25%; 2 + means CPE is 26%-50%; 3 + means CPE is 51%-75%; 4 + means CPE is 76%-100%). The inhibition rates of each drug on HCoV-229E in Huh7 cells were calculated by normalizing the CPE of each group to the CPE of virus-only well. The half toxic concentration (TC[50]) and the half inhibitory concentration (IC[50]) of each test sample were calculated using the Reed-Muench method. LibDock operation Molecular docking is a process that identifies the complementary molecules for a target spatially and electrically. The target protein structures were downloaded through the PDB protein database ([95]https://www.rcsb.org) [[96]41], and then imported into Discovery Studio 2016 software with small molecule structures. After protein and small molecule structure modification, LibDock was operated for molecular docking, with LibDock scores calculated as the assessment of molecular conformational affinity. Results Analysis of HCoV-associated targets of XCHD Previous preclinical studies and clinical trials demonstrated that the multi-component synergy of TCM are related to the interaction between components [[97]42, [98]43]. To analyze the chemical composition and pharmacodynamic material basis of XCHD, a total of 1,899 compounds in XCHD were obtained after removing the duplicates with identical chemical structures (Supplementary Table S[99]1). The numbers of ingredients of each herb in XCHD are 538 (Chaihu), 175 (Huangqin), 270 (Banxia), 472 (Shengjiang), 627 (Renshen), 18 (Zhigancao), and 253 (Dazao), respectively. To understand the potential synergistic effect mechanism of the herbs against HCoV, we used UpSet Wayne diagram to analyze the distribution of HCoV-associated genes regulated by XCHD (Fig. [100]2). There are a total of 37 HCov-associated targets regulated by all 7 herbs. In specific, the numbers of HCoV-associated genes targeted by the components of Chaihu, Huangqin, Banxia, Shengjiang, Renshen, Zhigancao and Dazao are 31, 11, 22, 16, 18, 1 and 30, respectively. Interestingly, inhibitor of kappa light polypeptide gene enhancer in B-cells, kinase beta (IKBKB) was the only common gene targeted by all seven herbs (Table [101]2), suggesting that IKBKB might be the most key target for the synergistic effect. In addition, we also observed there are 6 genes targeted by 6 herbs and 4 genes targeted by 5 herbs. The targets distribution reflects the potential mechanism of how XCHD performs synergistic effect against HCoVs. Fig. 2. [102]Fig. 2 [103]Open in a new tab HCoV-associated target distribution of various herbs in XCHD. Blue bars represent the number of targets for each herb, red bars represent the number of targets covered by single or multiple herbs, the dots indicate the targets associated by the ingredients in the corresponding herbs Table 2. Herbs in XCHD and their corresponding HCoV-associated host targets Herb Target gene symbol Chaihu ACE2, ANPEP, BCL2, BCL2A1, BCL2L1, CAV1, CD209, CHEK2, CLEC4M, CTSS, CXCL10, DPP4, FKBP1A, FURIN, G6PD, GBF1, HGS, IKBKB, IL6, IRF3, JUN, KPNA2, MCL1, PPIA, PPP1CA, PTBP1, SERPING1, SFTPD, STAT3, TGFB1, XPO1 Huangqin ACE2, BCL2, BCL2L1, CAV1, CHEK2, IKBKB, IL6, JUN, MCL1, STAT3, TGFB1 Banxia ANPEP, BCL2, BCL2A1, BCL2L1, CHEK2, DPP4, FURIN, G6PD, GBF1, HGS, IKBKB, IL6, IRF3, JUN, KPNA2, MCL1, PPIA, PPIG, PPP1CA, PTBP1, STAT3, TGFB1 Shengjiang BCL2, BCL2L1, CXCL10, DPP4, FKBP1A, FURIN, G6PD, IKBKB, IL6, IRF3, JUN, MCL1, PPIA, PSMA2, STAT3, TGFB1 Renshen ACE2, ANPEP, BCL2, CD209, CLEC4M, DPP4, FKBP1A, FURIN, IKBKB, IL6, JUN, MCL1, PPIA, SFTPD, STAT3, TERF1, TGFB1, UBE2I Zhigancao IKBKB Dazao ACE2, ANPEP, BCL2, BCL2A1, BCL2L1, CAV1, CD209, CHEK2, CLEC4M, COX2, CXCL10, DPP4, FKBP1A, FURIN, GBF1, HGS, IKBKB, IL6, IRF3, JUN, KPNA2, MCL1, PHB, PPIA, PPIG, PPP1CA, PTBP1, SFTPD, STAT3, TGFB1 [104]Open in a new tab Network construction and mechanisms analysis of XCHD against HCoVs In the study, to identify the effective components of XCHD against HCoV, we performed systems pharmacology-based framework analysis and obtained 344 components in XCHD connecting to 2,656 known targets and 561 predicated targets (Supplementary Table S[105]2 and Table S[106]3). We thus obtained 2,823 potential protein targets of XCHD by merging the known and predicted ones (Supplementary Table S[107]4). Furthermore, a global H-C-T network was constructed by integrating herb-compound pairs and compound-target interactions (CTIs), which is composed of 4,729 nodes (7 herbs, 1,899 compounds, and 2,823 protein targets) and 47,587 edges (24,545 herb-compound pairs and 45,133 CTIs). As illustrated in Fig. [108]3A, compounds and targets with degrees (D) larger than 20 were displayed. Most compounds were connected to multiple shared targets. Of note, the H-C-T network also indicated that several important HCoVs associated genes with high degrees (D > 20), including DPP4 (D = 59), BCL2 (D = 32), IL6 (D = 30), JUN (D = 25) and MCL1 (D = 23). Overall, it is likely that the ingredients in XCHD work to prevent HCoVs infection by regulating multiple HCoV-associated targets. Fig. 3. [109]Fig. 3 [110]Open in a new tab Network construction of XCHD against HCoVs. A A global herb-compound-target (H-C-T) network for XCHD. For demonstration purposes, only nodes with degree larger than 20 are displayed. The labels of the top 20 targets and compounds with highest degrees are displayed. B A specific compound-target (C-T) network of XCHD against HCoVs. The node size is proportional to degree. Chemical scaffold clustering analysis of the 161 XCHD constituents targeting to HCoV-associated genes (C) and the center chemical structures (D) In our study, 90 HCoV-associated targets were extracted from pharmacological references (Supplementary Table S[111]5). These genes