Research on anti-pancreatic cancer mechanism of Codonopsis codonopsis based on network pharmacology

Xiao-Qing Xu, Ya-Ping Yu, Bing-Shu Wang, Yong-Hao Fan, Wei Jie,2?, Shao-Jiang Zheng,2?

1. Tumor Institute of the First Affiliated Hospital,Hainan Medical University, Haikou 571199, China

2. Key Laboratory of Emergency and Trauma of Ministry of Education, Hainan Medical University, Haikou 571199, China

Keywords:Network pharmacology Codonopsis pilosula Pancreatic cancer Signal pathway

A BSTRACT Objective: The mechanism of Dangshen (Codonopsis pilosula) in treating pancreatic cancer(PC) was explored by network pharmacology technology and platform. Methods: The traditional Chinese medicine systems pharmacology database and analysis platform(TCMSP)was used to collect the effective compounds and potential targets of C.pilosula , and the genes associated with PC were obtained through the GeneCards database, the interaction genes between the effective compound targets of C.pilosula and PC targets were explored by the Venny method. The following mapping the interaction genes into a protein–protein interaction(PPI) network, and the key targets were screened. Finally, the interactive genes were imported into the DAVID database for gene ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) signal enrichment. Results: Twenty-one effective compounds and 98 downstream target genes of C.pilosula were screened through the TCMSP database. A total of 1,278 PC target genes were obtained through the GeneCards database, and the number of overlap genes between C.pilosula targets and PC related genes was 54, of which 10 were key node genes, namely CASP3, TP53, MDM2, AKT1, ESR1, BCL2L1, MCL1, HSP90AA1,CASP9, and CCND. These interactive genes involved a total of 30 typical GO terms and 20 KEGG signals. Conclusion: C. pilosula may play a role in treating PC through multicomponent, multi-target, and multi-signal pathways.

1. Introduction

Pancreatic cancer (PC) is a malignant digestive tumor with a poor prognosis. Despite decades of efforts, the five-year survival rate for PC is still less than 9%. Most patients are defined in the advantaged stage when they are first diagnosed [1]. The morbidity of PC has been rising rapidly around the world in recent years. Reports showed that the incidence of PC in China reached 12.17/100,000 in 2006, and it is expected that the mortality will move up to second place among all malignant tumors by 2030 [2]. Therefore, the prevention and treatment of PC have become one of the focuses of tumor research. However, the specific diagnostic biomarkers and therapeutic targets for PC are still lacking.

It has been reported that some traditional Chinese medicines,including Dangshen (Codonopsis pilosula), showed effects on the treatment of PC. C. pilosula is a perennial herb that contains multiple active ingredients like saponins, polysaccharides, and sterols. An increasing number of studies have shown the important roles of C.pilosula in anti-inflammatory[3], anti-ulcer [4], immune regulation[5-6], and promoting bone marrow hematopoiesis [7]. Meanwhile,the anti-cancer properties and therapeutic effects of C. pilosula have also been discovered in gastric cancer [10], colorectal cancer [11-12],melanoma [13], osteosarcoma [14], and other tumors. Particularly,Luan et al. found that C. pilosula could inhibit the proliferation of PC by promoting cell apoptosis and inhibiting cell proliferation [15].However, the molecular regulation mechanism behind the treatment of C. pilosula in PC remains unclear.

Network pharmacology, a computational method that integrates biological network and drug targets, has received great attention in exploring the pharmacological mechanism of drug treatment on diseases [16-17]. In this study, we predicted the active compounds of C. pilosula and their potential targets for PC through utilizing the network pharmacology technology and platform. Our findings provide a basis for understanding the mechanism of PC therapy based on C. pilosula.

2. Materials and Methods

2.1 Identification of effective compounds of C. pilosula and potential targets for PC

The traditional Chinese medicine systems pharmacology database and analysis platform (TCMSP, http://lsp.nwu.edu.cn/tcmsp.php) is a comprehensive Chinese herb resource to explore the relationship between Chinese medicine components and disease targets [18]. We collected effective compounds for C. pilosula under the threshold of oral bioavailability (OB) ≥30% and drug-likeness (DL) ≥ 0.18 and identified their corresponding targets. The obtained gene identities(IDs) were transformed into official gene symbols using STRING(https://string-db.org/).

2.2 Screening of PC-related genes

GeneCards (https://www.genecards.org/) is an intelligent retrieval system containing detailed annotation information on genes approved by the Human Genome Organization/Genome Database(HUGO/GDB) [19]. The database integrates genome, transcription,proteome, genetic, and functional information for all known and expected human genes. According to Zhang’s report [20], we screened PC-related genes through GeneCards with a score> 1.

2.3 Construction of the compound-target network for C.pilosula

The overlap gene lists between the targets of C. pilosula effective compounds and PC-related genes was obtained through Venny 2.1.0 (https://bioinfogp.cnb.csic.es/tools/venny/index.html). The compound-target network was constructed and visualized by Cytoscape 3.6.0, which was developed by JAVA and specifically analyzed a complex biological network. It can be used to depict network structure and hierarchy content, such as gene expression or protein–protein interaction (PPI) network [21].

2.4 Construction of PPI networks

We entered the overlap gene lists into the 'Multiple proteins' module of STRING to explore the related PPI networks. The interactions with confidence scores of more than 0.4 were retained and further visualized by Cytoscape. The cytoHubba plug-in was applied to identify network module and hub nodes. The top 10 genes were defined as hub nodes for C. pilosula treatment according to their network degrees. The nodes with higher degrees were colored red,while those with lower degrees were colored yellow.

2.5 Functional enrichment analysis

DAVID (https://david.ncifcrf.gov/) is a comprehensive set of functional annotation tools for investigators to understand the biological meaning behind a large list of genes [22]. We inputted the overlap gene lists to perform functional enrichment analysis,including GO and Kyoto Encyclopedia of Genes and Genomes(KEGG). We obtained the top 10 GO terms and 20 KEGG pathways with P-value < 0.05.

2.6 Statistical analysis

Data of each group were processed by the software package of its corresponding platform. P<0.05 was considered to indicate a statically significant result.

3. Results

3.1 The effective compounds of C. pilosula

We used the TCMSP to explore the effective compounds of C.pilosula. According to the setting OB and DL values as a threshold,we identified 21 effective compounds in total. The detailed information for these compounds is shown in Table 1.

Table 1 Information on 21 effective compounds of C. pilosula.

3.2 Venny analysis for the targets of effective compounds and PC-related genes

The predicted targets of C. pilosula effective compounds were obtained from TCMSP and further transformed into an official gene symbol by STRING. In total, we identified 98 genes that may be the targets of C. pilosula. In addition, through entering 'pancreatic cancer'into GeneCards, we collected 1,278 PC-related genes with a Score>1.Furthermore, 54 overlap genes between the targets of C. pilosula and PC-related genes were yielded by Venny analysis (Figure 1).

Figure1 Venny analysis oftargets for C. pilosula effective compounds and pancreatic cancer target genes.DS: Codonopsis pilosula; PC: pancreatic cancer

3.3 Compound-gene networks for C. pilosula

The compound-gene networks for C. pilosula were conducted using Cytoscape. The effective compounds connect to many overlap genes in the network, including luteolin, spinasterol, and glycitein (Figure 2). Therefore, these effective compounds may provide an important material basis for C. pilosula to exert its anti-cancer effect.

Figure2 Network diagram of "Codonopsis pilosula Compound-Interactive Gene."Red represents PC-related targets, and green is the active compound of Codonopsis pilosula.

3.4 C. pilosula pilosularelated PPI network in PC

A total of 54 overlap genes were passed into the “Multiple proteins”module and obtained the corresponding interactions (confidence score>0.4) from STRING. The PPI network was illustrated using Cytoscape software (Figure 3). Using cytoHubba module analysis,we obtained ten hub genes in the network, namely CASP3, TP53,MDM2, AKT1, ESR1, BCL2L1, MCL1, HSP90AA1, CASP9, and CCND1. The interactions of these hub genes are shown in Figure 4.

Figure3 The interaction network of C.pilosula pilosula and targets for pancreatic cancer.

Figure4 The interaction network of C. pilosula and the key proteins of PC targets.

3.5 GO and KEGG enrichment analysis

We applied DAVID to perform functional enrichment analysis based on the 54 overlap genes. The GO enrichment results,including biological process (BP), cellular component (CC), and molecular function (MF), are shown in Table 2. The most significant GO terms were GO:0007265～Ras protein signal transduction,GO:0005654～nucleoplasm, and GO:0019903～protein phosphatase binding for BP, CC, and MF, respectively. The KEGG enrichment results are shown in Table 3, hsa04660:T cell receptor signaling pathway, hsa04668:TNF signaling pathway, and hsa04115:p53 signaling pathway were the top three ranked pathways. Particularly,the detailed information about overlap genes in the hsa05212:PC pathway is shown in Figure 5.

Table 2 GO enrichment analysis

Table3 TOP 20 KEGG signaling pathways

Figure5 Enriched hsa05212: PC pathway The green proteins are the key molecules in enriched hsa05212: PC pathway.

4. Discussion

PC is a malignant tumor occurring in the digestive system, and its clinical diagnosis and treatment are extremely challenging [23].Chemotherapy is the main treatment for advanced PC. With the development of traditional Chinese pharmacology, utilizing Chinese medicine to treat tumors has received more attention. It has been reported that C. pilosula played pharmacological effects on human circulation, immunity, digestion, endocrine, and even in anti-tumor fields. Treatment with C. pilosula could be anti-mutation, improve the immune function, prevent tumor formation, and improve the efficacy of chemotherapy [24-26]. However, the reports mentioned above have failed to clarify the complex network regulation mechanisms of C.pilosula’s anti-tumor effect.

In this study, we first used the network pharmacology method and successfully screened out 21 effective compounds of C.pilosula. Among them, luteolin, spinasterol, and glycitein are the most important components with suitable OB and DL values.Luteolin globally exists in plants like vegetables, fruits, and Chinese medicines. It is a flavonoid compound that can confront human malignant tumors such as colon cancer, lung cancer, and PC. Luteolin could activate cell cycle arrest, inhibit tumor cell proliferation, and induce cell apoptosis. Moreover, it can prevent cancer development both in vitro and in vivo[27]. Spinasterol is a plant sterol with a structure similar to vitamin D, which exists in various plants. It is a key component of plant cells and can also be converted into equivalent prosterol glycosides. A study has reported that spinsterol isolated from Ganoderma lucidum had an effective inhibitory effect on breast cancer and ovarian cancer cell lines in a time and dose-dependent manner [28]. Glycitein, a kind of isoflavone,is involved in the reactive oxygen species (ROS)-related MAPK/STAT3/NF-κB signaling pathway. Glycitein could arrest the cell cycle at G0/G1 phase and induce apoptosis of gastric cancer AGS cells, thereby displaying the profiles of anti-gastric cancer [29].Apart from these three main ingredients, there are other 18 effective compounds of C. pilosula that deserve further exploring.

We next screened out 98 target genes of C. pilosula through STRING and GeneCards database, which may indicate the pharmacological multi-target characteristics of C. pilosula.Meanwhile, we also obtained 1,278 PC-related genes, which suggested the complexity of pancreatic tumorigenesis. Through Venny analysis, 54 overlapping genes between the targets of C. pilosula and PC-related genes were identified. The layout of constructed compound-gene network demonstrated the multi-target characteristics of C. pilosula in the treatment of PC. The ten hub nodes in this network were ESR1, HSP90AA1, CASP9, CASP3,TP53, MDM2, AKT1, BCL2L1, MCL1, and CCND1.These results provided potential targets for the treatment of PC with C. pilosula in the future.

Estrogen receptor 1 (ESR1) is a transcription factor that promotes cell survival and proliferation. ESR1 was found consistently expressed in about 70% of breast cancer patients and to mediate the tumor drug resistance [30]. In addition, ESR1 could interact with multiple protein kinases, thus formatting protein complexes and stimulating the downstream molecules such as Akt in the process of estrogen signal transduction [31]. The potential role of ESR1 in pancreatic tumors has been debated for many years; recent studies have shown that ER expression is associated with a poor prognosis for PC patients [32-33].

Heat shock protein 90 alpha family class A member 1 (HSP90AA1,also known as Hsp90α) could interact with proteins involving carcinogenesis, extracellular wound healing, and inflammation.The overexpression of Hsp90α in PC cells is related to the poor prognosis. Related studies have shown that the expression level of Hsp90α in the cytoplasm of PC cells is significantly higher than in normal pancreatic tissues. Still, the expression level of Hsp90α in the nucleus of PC cells is lower. In addition, the expression of Hsp90α in the cytoplasm of PC cells is significantly associated with peripheral nerve invasion. [34].

Tumor protein p53 (TP53) is a tumor suppressor gene whose product can regulate cell growth, aging, and tumor process. TP53 is often regarded as the guardian of the human genome. There are a large number of mutations in the TP53 gene region for PC patients[35]. Moreover, researchers have identified SERPINE1 as a target of the TP53/miR-34a axis, which may be a potential biomarker for early PC detection [36].

The amplification of MDM2 proto-oncogene (MDM2) is closely related to tumor metastasis. Therefore, analyzing the alteration of MDM2 is very important for understanding PC's etiology and metastasis mechanism. The expression level of MDM2 was significantly increased with the degree of PC differentiation.Moreover, the expression of MDM2 in PC is significantly higher than in normal tissues [37].

The AKT serine/threonine kinase 1 (AKT1) encodes a serine/threonine-protein kinase, triggered by an extracellular signal pathway through phosphatidylinositol 3-kinase (PI3K). AKT1 would move to the nucleus, cytoplasm, and other cell parts to phosphorylate a variety of substrate proteins if activated. At the same time, the abnormal activity of AKT1 can cause tumors. Thus, the AKT pathway is the main regulator of human PC progression and an important pharmacological target. In preclinical models, the combination of AKT and in vivo pharmacology of mitochondrial metabolism could inhibit and effectively control the growth of PC [38].

BCL2 apoptosis regulator (BCL2) can inhibit cell apoptosis. BCL2 upregulation has been found in various human tumors, which is closely related to tumor occurrence, development, and treatment[39]. In PC, the increased expression of BCL2 plays a role in Rab14-mediated gemcitabine resistance [40]. In addition, several Chinese herbal medicines can target BCL2 to induce apoptosis in PC cells which highlights the value of BCL2 in PC therapy [41].

Myeloid cell leukemia-1 (MCL1) is a member of the BCL2 apoptosis-regulating gene family. Using RNAi to specifically target and silence the MCL1 in the human PC cell line PANC-1 could significantly induce cell apoptosis and reduce cell proliferation activity. These results suggested that the RNAi technology targeting MCL1 has potential value in gene therapy [42].

Cyclin D1 (CCND1) is found to be overexpressed in multiple cancer types [43]. CCND1 could function on the cyclin-dependent kinases CDK4 and CDK6 to regulate the transition of the cell cycle from the G1 phase to the S phase, thus improving cell growth and leading to cancer. Experiments have shown that CCND1 is the direct target gene of miR-584 in PC. In addition, overexpression of miR-584 could inhibit PC cells, which is consistent with the inhibitory effect of CCND1. Moreover, the restoration of CCND1 expression significantly eliminates the effect of miR-584 on the PC cells [44].

Caspase is acysteinyl aspartate-specific proteinase that regulates cellular differentiation, growth, and apoptosis, and determines the morphological and biological changes of apoptosis [45]. Under the stimulation of foreign protein signals, Caspase-9 cleaves and activates Caspase-3, and the activated Caspase-3 can cause programmed cell death. Long et al. [46] found that the expression of Caspase-3 in normal pancreatic tissue was lower than that in PC tissue. The experiment of Du et al. [47] suggested that the expression of Caspase-3 in normal pancreatic tissue was significantly higher than that in PC tissue. Pu et al. [48] reported that the expression level of Caspase-3 in PC was downregulated with the decrease of tumor differentiation. The above findings suggest that the members of the Caspase family play an important role in the clinical progress of PC, but the specific expression pattern and role need to be further clarified.

We performed GO and KEGG functional enrichment analysis based on the 54 overlapping genes. The GO:0007265～Ras protein signal transduction is the most significant GO-term in BP.Studies have shown that Ras mutation is highly related to PC[49-50]. GO:0005654～nucleoplasm is the most significant GO-term in CC, suggesting that C. pilosula related target genes are mainly located in the nucleoplasm and may be related to downstream signal transduction. GO:0019903～protein phosphatase binding is the most significant GO-term in MF, suggesting that C. pilosula related target genes have important protein modification functions.The KEGG pathway enrichment results found that the C. pilosula target genes were mainly enriched in the following aspects. First, C.pilosula was related to inflammation and immune regulation. The main KEGG involved in these processes includes hsa04620: Tolllike receptor signaling pathway, hsa04660:T cell receptor signaling pathway, hsa04668:TNF, and hsa04662:B cell receptor signaling pathway. The second is the signal transduction pathway, which includes the hsa04915:Estrogen signaling pathway, hsa04012:ErbB signaling pathway, hsa04630:Jak-STAT signaling pathway, and hsa04010:MAPK signaling pathway. The third is cell cycle and apoptosis regulation signals, which mainly involved hsa04115:p53 signaling pathway and hsa04110:Cell cycle pathway. The abnormalities of the C. pilosula related genes mentioned above are further reflected in the hsa05212: PC pathway. The complexity of the KEGG signaling pathway suggests that C. pilosula may have a multinetwork interaction/cooperation mechanism in the treatment of PC.

In conclusion, this study summarized the categories of effective C. pilosula, predicted the target genes of these effective compounds in PC, obtained key hub genes, and performed relevant function annotations and KEGG signal enrichment, providing a theoretical basis for the treatment of PC. Thus, this research integrates bioinformatics and pharmacology through network pharmacology technology. For the first time, it established a feasible strategy for developing a C. pilosula PC treatment based on big data analysis.However, the relevant results still need experimental verification.

Conflict of interest statement

All authors have no conflict of interest.

Author contribution

Experimental design: WJ, SJZ; Literature retrieval: XQX,YPY, BSW, YHF; Data statistics: XQX, YPY, BSW, YHF; Chart production: XQX, YPY, WJ; Manuscript writing: XQX, YPY, WJ;Fund acquisition: SJZ.