| JAK-STAT is an important cell signaling pathway involved in regulating the cell proliferation, differentiation and apoptosis. Most members express in adult liver and play important role in liver regeneration, immunnity, etc. What's more, abnormal signaling transduction of the JAK-STAT could cause liver cancer and metabolic syndrome. These facts suggest that the JAK-STAT signaling pathway is very important for the adult liver.The signal transduction is one of the hottest spot of life sciences, and many breakthroughs have been made, but there are still many important issues needing more thorough study, even the most basic issue, for example, how many proteins are actually involved in each signaling pathway? And what's the different constitutes of a signal transduction pathway in different cells or organs? Therefore, it is of great significance to construct the protein interaction network of JAK-STAT signaling pathway of the normal adult liver to understand the structural characteristics and its complexity, at the same time, the result is helpful to study the specific control laws and the molecular mechanism of liver function, which may leads to find new methods for liver disease therapy and new proteins for drug development.Protein interaction is critical for cell's function. And the construction and analysis of protein interaction network is an important method of systems biology to study protein functions, molecular and cellular mechanisms. Among the large scale methods to study protein interaction, yeast two-hybrid assay is an important method, which has been successfully applied to construct the interaction networks of yeast, nematode, fruit fly, and human. Therefore, we selected the JAK-STAT proteins expressed in the liver as baits, using the stringent yeast two-hybrid system to screen human liver cDNA library.42 genes were selected to construct 106 baits of different domain fragments or full length, and 98 baits of 41 genes were successfully constructed. After self-activating test, 20 baits could not be used for yeast two-hybrid screening for self-activation. So 78 baits of 31 genes were applied to liver cDNA library screening, and more than 3,000 positive clones were obtained. Of them, 1,480 clones were sent for sequencing and 1,384 clones were successfully sequenced. BLASTP resuls showed that 24 baits did not get any prey, but 54 baits of 22 genes obtained 1-63 preys, and finally 345 interaction were obtained between 22 genes and 262 preys, with an average of 15.6 interactions per gene (343/22), which was similar to the result of Janghoo Lim'report, 18.3 interactions per gene.Stringent yeast two-hybrid system was used in our study to decrease the false positive results, but it was impossible to get rid of all the false positives in large-scale screening. In order to estimate our technical false positive rate, 304 interactions were tested in retransformation assay, and 53.3% interactions showed positive phenotype. With these results, it should be reasoned that there existed technical false positives in our data set. But only 66.6% interactions repeating the positive phenotype for the known interactions seguested that prominent false negative rate lied in the retest assay. So, the negative interactions in the retest assay can't be removed, and other independent experiments should be employed to evaluate the technical false positive rate of our interactions.We employed 12 different bioinformatics methods to evaluate biological false positives, And found 32.5% (112/345) interactions had interaction domains; 1.7% (6/345) interactions had been verified in other species; 46.4%(160/345) interactions shared at least three level GO component annotation; 35.9% (124/345) interactions shared at least three level GO process or function annotation ; 11% (38/345) interactions showed correlated expression pattern in mRNA level; 32.4% (112/345) interactions formed three-protein-interaction loop with other proteins in the big networks; 56.8 % (196/345) interactions formed four-protein-interaction loop with other proteins in the big networks; 2% (7/345) interactions shared the common biological pathway; 29.3%(101/345)interactions showed the same or similar phenotypes in knockout mice; 2% (7/345) interactions caused the same or similar disease; and the prey of 2% (7/345) interactions were identified as the components of JAK-STAT in Drosophila melanogaster by large-scale RNAi screening. All of these imformation suggested that some of the interactions were of high confidence.To evaluate the confidence of the interactions, we integrated the experimental data and the bioinformatics data to establish a comprehensive score system, including 4 experimental and 12 bioinformatics items, Each interaction was scored by the 16 items, and the score of all the 16 items were added up, and then we grouped all the interactions into 3 confidence level. The interactions with low, medium and high confidence scored from 0.35 to 2.5, 2 to 5, and 5 to 14.6 respectively. And we found 50.1%(175/345) our interactions were of low confidence, 36.5% (126/345) medium, 12.8% (44/345) high. So about half of our interactions might be real. Finally, we presented the interactions in visible network graphs with Osprey network software. The networks showed that our dataset was complement for the HPRD curated networks. For the networks, the main shape did not change, but several baits with more reported interactions were more highly connected with more new preys in the integrated networks. For example, with STAT3, 58 new preys were added in the integrated network. Indicating that STAT3 was very important in JAK-STAT signaling pathway of the adult liver and its transcription activity was under tight control. In our interaction network, we found that the proteins with transcription and signal transduction function could connect into network separately with three-protein-interaction loop and four-protein-interaction loops,in which NR4A1 protein took the center stage and might play much more important role in human adult liver than what we had recognized. In addition, there were 65 preys and baits associated with human diseases recorded in OMIM, suggesting that our Y2H interaction networks and the integrated ones could be used to analyze the molecular mechanism of diseases and provided functional clues for unknown proteins. Also, new functions for known proteins might be found from the networks. Finally, by combining the OSPREY software visualization with the score system analysis and published references, we found several interactions might be of great importance: (1) STAT3-DVL1, potentially new crosstalk between JAK-STAT and Wnt signal transduction pathway; (2) STAT3-SIAH2, probably new E3 ligase SIAH2 for degradating STAT3 in the nucleus; (3) JAK2-CHD3, further confirming the role of JAK to change heterochromatin structure and to activate non-specific gene transcription; (4) STAT3-RUNX1, STAT3-NR4A1, STAT3-ZC3H12A, these new interactions showed the transcriptors directly connected to each other or form complex to coordinate different target gene transcription.In conclusion, our results present a primary network for JAK-STAT signaling transduction of human liver. The interactions obtained from this study will help to understand the complicated regulations of liver functions and the physiological and pathological processes with various liver disorders. At the same time, we first systematically construct the organ specific interaction network of signaling transduction pathway in human with yeast two-hybrid, which demonstrates the same strategy can be applied to study special signaling transduction networks of other human organs or tissues. |
PDF Full Text Request