| Eukaryotic cells include many dynamic membrane-bounded organelles that carry out distinct functions. Although these cell compartments have been studied for a long time, the diversity of proteins in different organelles remains unclear. With the development of mass spectrometry, sub-cellular proteome has become one of the most important fields of proteomics, which provide a powerful tool to give a survey of the proteins composition in different organelles. What's more, organelle proteomics is also a crucial part of Human Liver Proteome Project.Here, we combined sub-cellular fractionation with proteomic technique to explore the protein composition of liver plasma membrane. In addition, the separated organelles always contain cross-contaminant by other compartments and could be identified by high sensitive mass spectrometer easily, which may lead to miss-assessment of protein localization. To address this issue, we used a machine-learning stratry to assign protein subcellular localization. Based on this method, the datasets of liver plasma membrane proteins was confirmed, and the function of those proteins was analysis. More importantly, organelle proteomic maps provides a''cell biological scaffold''on which other functional genomics data can be layered. And the integration of many diverse data sets can help accquire a novel view of the organelles.Combined with protein-protein interaction data, protein models were assembled for annotation organelles and assigning function to uncharacterized proteins. Moreover, it is known to all that genome annotation by proteomic data is an important and difficult field. Here, combining mouse sub-cellular proteome and EST sequence database, we found new protein-code genes and novel gene models, which could be used as a preparation for the next step of human genome annotation. The detailed research ideas and main results outlined as follows.Firstly,we established a method to obtain multiple organelles from same liver homogenate. C57BL/6J mouse liver was chosen as a model to explore the optimum method for sub-cellular preparation. The method could obtain the multiple fractions including plasma membrane, mitochondria, nucleus, rough and smooth endoplasmic reticulum and cytosol from a single homogenate. We systematically evaluated the purity, efficiency and integrity by western blot and transmission electron microscope. Subsequently, highly purified plasma membranes from human liver were used to compile a protein expression profile with strategy that combined SDS-PAGE separation with liquid chromatography gas phase fractionation (GPF) tandem mass spectrometry analysis. Totally, 1381 human proteins were identified with 95% confidence and minimum two peptides match. And the quantitation of proteins was obtained by the number of mass spectra.Secondly, the sub-cellular localization of human liver protein was analyzed. We developed a machine-learning method named KNN (K Nearest Neighbor) which was based on the similarity of quantitative curve in different compartment comparing to organelle marker protein. In addition, Bayesian model was employed, which combined KNN with five sub-cellular localization prediction methods (pTAGET,Proteome Analyst,WoLFPSORT,TargetP and NUCLEO), to assign the sub-cellular localization of human liver proteins. Comparing with the well-known organelles marker proteins, we found 4966 new sub-cellular localizations, including two categories: one was that the proteins'localization was never found before, and the other was that new localization was given to the well-reported proteins. At last, four new protein localizations were selected and confirmed by experiments.Thirdly, the function of human liver plasma membrane was analyzed. Based on Bayes model, a total of 871 proteins were found, 84% of which were firstly given the localization to plasma membrane. Except the single localization, the plasma membrane and endoplasmic reticulum-targeted proteins had the largest proportion (215, 24.68%). Those protiens were mainly involved in the protein secretory pathway, which reflect the strong secretion function of the liver. In the function analysis, the biological characterization was described systematically in liver plasma membrane, in which many proteins involve in signal transduction, ion transport and protein modifing. Further more, we found 29 the hypothetical proteins, of which 10 proteins had both C1-set and V-set domains. By combining with sub-cellular localization and domain information, we predicted those proteins may belong to the immune globulin superfamily, which involved in the process of immune response.Fourthly, the function modules of different organelle were constructed for predicting of protein function. The aim of organelle proteome research is not only to provide an organelle reference map, but also to form new knowledge. Here, we combined the human liver sub-cellular proteome with the protein interaction network to acquire the organelle specific protein network, which was clustered into function modules by MCODE. Based on those modules, the relations of organelle localization, protein quantitative and function were explored. Importantly, the members of protein module had similar function, which could be used as indicator for unknown member. Based on this principle, the new function of 151 proteins were predicted, one of which was a novel protein involving in mitochondria Respiratory chain complex I. We had demonstrated that the localization data presented here was not only valuable by itself but could be combined with other large-scale datasets to gain unanticipated insights.Finally, the mouse genome was annotated by mouse liver proteomic data. Genome annotation was an important field of proteomic study, by which novel gene and gene mode could be found. By searching mouse EST sequence database and blasting with mouse genome, 486 novel peptides were obtained including 96 novel protein-coding genes 152 novel amino acids mutations,102 novel peptides matching no genome sequence and 136 new peptides never existing in IPI database.In conclusion, we established a method that could separate multiple fractions from a single liver homogenate and identified 1381 human liver plasma membrane proteins. After that, a new sub-cellular localization strategy was used, and 4966 new localizations was found,four of which were confirmed by experiments. Then, the function of plasma membrane proteins was analysis, and 10 hypothetical plasma membrane proteins may belong to the immune globulin superfamily. Subsequently, the function modules of different organelle were constructed to predict 151 new proteins function. At last, 486 novel peptides, including 96 novel protein-coding genes, were found by searching the mouse EST database. |
PDF Full Text Request