Identify And Compare The Core Module From Network Of Protein-Protein Interation After Different Compound Of Qingkailing Treatments On Cerebral Ischemia Models

BackgroundCerebral ischemia is considered to be induced by the interaction of multiple pathways, which is a complex disease involving a series of biochemical and molecular mechanisms with multiple genes, pathways and targets. Systems analysis of drug intervention remained scarce."Bioactivity-guided chemistry research ideas," proposed by Wang yongyan emphasizes new formula composing principles base on effective components of traditional Chinese medicine, aiming at clarifying the drug components, specific targets and composing principles. Although the complexity of disease is a thorny issue, fortunately, network analysis as a new tool can be used for the integration of the complex relationship between the disease and drugs. Evidence showed that module structure is a key factor for understanding the biological systems. Module pharmacology considers that the treatment of complex diseases drives from/requires modular intervention/design to affect multiple targets. Confirming the core module from complex networks will help to understand the network behavior and predict future dynamic characteristics.Experiment shows that the effective components such as baicalin, jasminoidin and ursodeoxycholic acid extracted from Qingkailing can effectively reduce infarct volume in MCAO model. Based on the theoretical framework of module pharmacology, this study will further deepen the understanding on pharmacological mechanism of these effective components mentioned above. ObjectiveThe purpose is to identify the core module from the protein-protein interactive networks which built by the whole gene expression data that from the intervention on MCAO by using those effective components.MethodsGenome-wide analysis of the data by the IPA database, wherein the gene set with functional statistical significance was used for network mapping. Single-scale and multi-scale network skeleton as a background were extracted from STRING database for constructing the weighed networks. MCODE and Pyramabs were used for dividing modules and these modules were used to construct the weighed module networks. By using node strength, betweenness centrality and Page rank three computational methods for screening candidate core module from module networks, then, utilizing network characteristic path length determined core module finally. DAVID database was selected for GO functional annotation for the core module. Protein level verification used westen blot.Results1Single scale network construction, Gvehicle:N=2853, E=13117; GBA:N=2572, E=14187; GJA:G=3656, E=16617; GUA:N=3059, E=16878; A power-law distribution network, the range of node weights:0.9<W<1. The construction of multi-scale network, the SigScore is0.8determined by modularity calculation and target map scale, GVehicle:N=3750, E=9162; GBA:N=2813, E=6217; GJA:N=3416, E=7581;:N=3407> E=9057; A power-law distribution network, the range of node weights:0<W<1.2In the non-hierarchical module partition method, MCODE was selected by using calculate minimum entropy of network module. 3Results of module partition3.1Under MCODE method, Vehicle, BA, JA, UA group was divided into37,21,25,26module in unweiighted single scale network, respectively, and were30,24,24,19module in unweiighted multi-scale network, respectively.3.2Under Pyramabs method, in unweiighted single scale network, module number of high-level and low-level in Vehicle, BA, JA, UA group were20,75,9,39,13,74,14,75, respectively. In weiighted single scale network, module number of high-level and low-level in Vehicle, BA, JA, UA group were10,42,7,41,20,110,10,96, respectively.In unweiighted multi-scale network, module number of high-level and low-level in Vehicle, BA, JA, UA group were35,175,19,93,31,142,23,113, respectively. In weiighted single scale network, module number of high-level and low-level in Vehicle, BA, JA, UA group were20,89,22,90,24,150,21,115, respectively.4The core module in Vehicle, BA, JA, UA group were14,15,10,12, respectively. The first module detection between betweenness centrality and Pagerank were convergence, the proportion of check out was82.5%. The pair core module which have a inclusion relationship between lower level and high level, illustrate centricity position up or down, rather than the un-inclusion relation is reflected the transfer between the position in the hierarchy. No matter the type of network scale or the method of module identification is different in every group, the component between almost all of main module have overlapping parts.5Each core module of model group and drug group, GO enrichment overall function is associated with cerebral ischemia. In the core module,1457,1408,1342,1228biology function and100,111,86,85KEGG pathway were enrichment of Vehicle, BA, JA, UA group, respectively, they were both overlap and non-overlap. 6The results of Western blot showed that, compared with the sham group, the protein expression levels of PDZK1was significantly up-regulated in vehicle group (P<0.05). Compared with the vehicle group, PDZK1protein expression levels were significantly down-regulated by BA and JA (P<0.05).Conclusion1For the protein interaction network of cerebral ischemia model and drug intervention with BA, JA and UA that to identify the core module by using of node strength, betweenness centrality and page rank is feasible.2To identification the core module under the protein network background with different scale and weights is innovated. The main module in each group are polymorphic.3For model group the core module in drug group show the tendency of centralized and integration regulation both in the biological function and signal pathway.4Module network construction and the core module identification for understanding complex diseases after drug intervention and complex pharmacological mechanism of combination therapy provide efficient analysis strategy.