MitoP2: the mitochondrial proteome database—now including mouse data
http://www.100md.com
《核酸研究医学期刊》
1Institute of Human Genetics, Technical University of Munich Munich, Germany 2Institute of Human Genetics, GSF National Research Center for Environment and Health Neuherberg, Germany 3Institute for Bioinformatics, GSF National Research Center for Environment and Health Neuherberg, Germany 4Department of Biochemistry and Stanford Genome Technology Center 855 California Avenue, Palo Alto, CA 94304, USA
*To whom correspondence should be addressed. Tel: +49 89 3187 2890; Fax: +49 89 3187 3297; Email: prokisch@gsf.de
ABSTRACT
The MitoP2 database (http://www.mitop.de) integrates information on mitochondrial proteins, their molecular functions and associated diseases. The central database features are manually annotated reference proteins localized or functionally associated with mitochondria supplied for yeast, human and mouse. MitoP2 enables (i) the identification of putative orthologous proteins between these species to study evolutionarily conserved functions and pathways; (ii) the integration of data from systematic genome-wide studies such as proteomics and deletion phenotype screening; (iii) the prediction of novel mitochondrial proteins using data integration and the assignment of evidence scores; and (iv) systematic searches that aim to find the genes that underlie common and rare mitochondrial diseases. The data and analysis files are referenced to data sources in PubMed and other online databases and can be easily downloaded. MitoP2 users can explore the relationship between mitochondrial dysfunctions and disease and utilize this information to conduct systems biology approaches on mitochondria.
INTRODUCTION
The application of genomics to biology and medicine requires an understanding how specific gene variants contribute to phenotypes, in combination with a comprehensive knowledge of the ‘parts list’ of a cellular system and how these components are assembled into functional units (1). Mitochondria are ubiquitous and defined substructures of nucleated cells and lend themselves to systems biology approaches. However, in generic databases the annotation of mitochondrial proteins is often incomplete and does not always distinguish between proteins which have a confirmed mitochondrial subcellular localization and those which are only candidates according to preliminary experimental results or in silico predictions. For the human species, about half of the estimated 1500 proteins localized or functionally associated with mitochondria are known (2). Since the mitochondrial organelle is an evolutionarily conserved entity, systematic studies in model organisms are powerful to identify mitochondrial proteins in other organisms (3).
The MitoP2 database was created to consolidate and structure public information on mitochondrial proteins, their functions and associated human diseases (4,5). MitoP2 provides a wide variety of search functions to explore and download information and to access references in PubMed and other public databases. We have further expanded the manually annotated reference sets of mitochondrial proteins in yeast (522 proteins) and human (624 proteins), and have now added the section MitoP2-Mouse (615 proteins). For these three species, we integrated data from genome-wide approaches applied to the study of mitochondria, and assigned an evidence score of a candidate protein being mitochondrial (3). With the help of MitoP2, proteins involved in mitochondrial biogenesis and function have been identified and characterized (6,7). In addition, MitoP2 has enabled the identification of disease genes using positional candidate approaches (8–10).
MitoP2-YEAST
A wealth of information has been collected over the past several years from single gene and genome-wide studies of Saccharomyces cerevisiae (11). The list of yeast ORFs and protein annotations in MitoP2-Yeast are based on information in the Saccharomyces Genome Database (SGD; http://www.yeastgenome.org) (12). This MitoP2-Yeast update now provides annotated information for 522 mitochondrial reference proteins, which are based on experimental validation of each of these proteins. Recently, systematic cellular sublocalization studies estimated a total of 800 mitochondrial proteins presenting 12% of the currently known yeast genes (13,14). Therefore, 250–300 mitochondrial proteins are still missing. In order to identify these missing genes, we have validated and integrated genome-wide approaches applied to the study of mitochondria (3). MitoP2-Yeast datasets in Table 1 show 20 systematic approaches used for this purpose: phenotypes of single gene deletion mutant phenotypes (15,16); systematic subcellular localization studies (13,14); transcriptome datasets of differentially expressed genes including fermentable and non-fermentable growth conditions, the response to diauxic shift, and Hap4 transcription factor screening (3,17,18); proteome analyses of purified mitochondrial organelles (3,19,20); protein abundance measurements (21); and data from protein–protein interaction studies that include interactions to mitochondrial proteins (22). In addition to experimental datasets, mitochondrial proteins can be predicted in silico based on the presence of mitochondrial targeting sequences (23–26), and by sequence similarity to a known mitochondrial protein from other species (defined as bidirectional best BLAST hit or best BLAST hit with a score <1 x 10–10) (27). Data from each of these systematic studies can be searched and downloaded.
Table 1 Comparison of specificity and sensitivity for various approaches integrated in MitoP2 in determining the mitochondrial localization of proteins
Using the MitoP2-Yeast reference proteins, it is possible to analyze the specificity and sensitivity of the data from genome-wide studies (Figure 1). Specificity is defined as the proportion of proteins of a dataset which are part of the reference set, while sensitivity is the proportion of reference set proteins which is covered by the dataset. In order to identify putative mitochondrial proteins, we calculated a MitoP2 score for each protein, reflecting the specificity of combined approaches which identified the particular protein (3). To further improve these predictions, we used a new approach utilizing a support vector machine (SVM, http://svmlight.joachims.org). The SVMs are learning machines based on statistical learning theory used for solving classification tasks. We trained the SVM using the MitoP2-Yeast reference set (522 proteins) and a set of 519 proteins with a known localization to other cellular compartments collected from SGD (http://www.yeastgenome.org/). For each of the 1041 proteins, we defined a 20-dimensional vector using the datasets of 20 systematic studies (see Table 1). This resulted in a 20-dimensional input matrix, which was used to train the SVM (see also Supplementary Figure S1). After training, the SVM predicts mitochondrial proteins with a specificity of 78% and a sensitivity of 80% (SVM score >1). This analysis shows that a combination of datasets from genome-wide studies significantly increases the power of predicting mitochondrial proteins beyond the level achieved by any single study (Figure 1).
Figure 1 Systematic approaches to identify mitochondrial proteins. The yeast datasets were benchmarked against the mitochondrial reference set. Each point represents a dataset whose position is determined by benchmarking against the 522 reference proteins from MitoP2-Yeast. The different groups of approaches are highlighted using distinct colours: the bioinformatics datasets (purple) are PSORT (25), MitoProt >90 (23), Bayesian prediction (37), Predotar (26) and yeast proteins with human mitochondrial orthologs (MitoP2 database); the experimental datasets (blue) are as follows: hap4 expression (18), respiration induced expression (3), mitochondria localized ribosomes (38), deletion phenotype screen (16), tag localization (14), GFP localization (13), pet phenotypes (15), four mass spectrometry proteome studies (3,19,20,39) and high and medium confidence protein–protein interactions (PPI) (22) defined by interactions with known mitochondrial proteins (MitoP2 database). The predictive score for a mitochondrial protein (MitoP2 score; green) was based on the combination of the systematic datasets, calculated for different thresholds. The predictions using the SVM algorithm are shown in red for different thresholds.
MitoP2-HUMAN AND MitoP2-MOUSE
We manually annotated mitochondrial reference proteins for human (624) and mouse (615) that now cover about half of the estimated mitochondrial proteins in these two species. These reference proteins present a subset of all the protein entries in the database: MitoP2-Human contains 36 504 proteins and MitoP2-Mouse contains 32 422 proteins. These datasets have been downloaded from the Swiss-Prot database (http://www.expasy.org/sprot/) (28). To identify putative orthologue proteins between human and mouse we calculated a bidirectional best BLAST hit or a best BLAST hit with score <1 x 10–10 between the two datasets. For each MitoP2 protein, we extracted descriptions, chromosomal positions, subcellular localization and literature references from Swiss-Prot. In addition, functional annotations such as biological processes and functional categories were extracted from the Gene Ontology database (GO; http://www.geneontology.org/). For MitoP2-Mouse, we annotated functional descriptions according to the MIPS functional catalogue (29), and provided access to DNA and protein sequence information. Each of these protein annotations is accompanied by its PubMed reference link. Phenotypic information on available mouse models are provided by the Mouse Genome Informatics database (MGI; http://www.informatics.jax.org/) (30). To date, more than 50 mouse models carrying mutations or deletions of mitochondrial genes have been investigated. For researchers interested in studying these models, MitoP2 provides links to the International Gene Trap Consortium (IGTC; http://www.genetrap.org/) to access the related mouse cell lines.
MitoP2-Mouse and MitoP2-Human provide similar search options that allow single or combined searches for individual database components (Figure 2). Database searches and downloads can be performed using keywords, genes names and the selection of datasets from systematic studies. For MitoP2-Human, two proteome studies on mitochondrial organelles purified from heart tissue are available, which have been integrated under the ‘proteome’ category (31,32). For Mitop2-Mouse, we integrated the datasets from three high-throughput studies that include two proteome experiments (33,34) and a subcellular localization study using split-enhanced green fluorescent protein (EGFP) (35). For mouse proteins identified using these approaches, we identified the respective putative orthologue proteins in human, and vice versa. The number of proteins in human and mouse datasets differ in part due to missing proteins in either one of the species in Swiss-Prot. The MitoP2 category ‘transcriptome’ predicts gene relationships based on similarities of their expression profiles (34). In addition to sequence similarity searches between human, yeast and mouse, MitoP2 provides in silico predictions for mitochondrial proteins utilizing established algorithms such as MitoProt II (23), PSORT II (25), Predotar (26) and MITOPRED (24). These programs allow the prediction of subcellular localizations of proteins based on their amino acid sequences. To illustrate the different search functions, users can select PSORT II under MitoP2-Mouse to extract 4321 proteins that include 323 entries from the mitochondrial reference set (7%). Alternatively, one can perform combined searches, for example, by selecting PSORT II and a human proteome dataset ‘Hprot_01’ (31,32), which then generates a list of 176 proteins that includes 56% of the mitochondrial reference set. This comparison demonstrates the trade-off between sensitivity and specificity: the combination of datasets reduces the total number of proteins (sensitivity), while it increases the specificity for mitochondrial proteins.
Figure 2 Screenshot of the MitoP2-Mouse query page. The MitoP2 query page is structured according to various groups of search parameters provided by the database. The search options are either linked to the online references or an explanation for this selection is provided.
Each entry in MitoP2-Human and MitoP2-Mouse corresponds to a Swiss-Prot identifier with protein descriptions, annotated subcellular localization and sequence map positions according to UCSC genome browser (http://genome.ucsc.edu/). In addition, the single protein entry summarizes the information from in silico predictions, high-throughput experiments, the availability of mouse gene trap clones and the predictive MitoP2 score. An example for a single protein entry in MitoP2-Mouse, the adenine nucleotide (ADP/ATP) translocator 2, is shown in Figure 3. This figure shows in a matrix lane, the information available for this protein extracted from Swiss-Prot and the integrated genome-wide approaches, a list of functional annotations compiled from the MIPS catalogue that are linked to the Mouse Functional Genome Database (http://mips.gsf.de/genre/proj/mfungd/), PubMed links to the references, and a list of similar sequences from other species. The other parts of this entry, which are not shown in this figure, include a phenotype description of the associated mouse mutant, the Gene Ontology annotations for molecular protein functions and biological processes, literature references on protein functions and variants that are listed by author names and title and a table of Swiss-Prot references.
Figure 3 Example for protein entry in MitoP2-Mouse. As illustrated for the mitochondrial ADP/ATP carrier protein 2 (ADT2). MitoP2 provides for each protein entry the Swiss-Prot name and description, the chromosomal localization, results from mitochondrial prediction programs, data from proteome studies, available gene trap clones, functional annotations according to MIPS, PubMed reference links and homologous proteins in other species.
For genes implicated in a hereditary disease, MitoP2 provides a link to the corresponding entry in the Online Mendelian Inheritance in Man database (OMIM; http://www.ncbi.nlm.nih.gov) (36). To date, more than 120 of the 624 human mitochondrial proteins are known to be involved in a hereditary disease. Mitochondrial disorders have a diversity of debilitating phenotypes and include a wide variety of neurodegenerative processes, cardiovascular disorders, diabetes mellitus and several cancer types. Many of these disease genes function in the metabolism of amino acids, nucleic acid, fatty acids and lipids, and energy production. The MitoP2 database enables the systematic identification of candidate genes to study mitochondrial diseases (5). Elpeleg et al. (8), for example, mapped a locus for hereditary mtDNA depletions associated with mitochondrial encephalomyopathy to a 21 Mb interval on chromosome 13. The mapping coordinates (i.e. 13:40878920 and 13:61359487) were used as a selection criteria to prioritize MitoP2 candidate genes among the 113 genes predicted in this region. In combination with a MitoP2 score >60, three proteins were identified as disease candidate genes. One of these genes (SUCLA2), a mitochondrial reference protein identified in two proteome experiments, was found to be mutated in affected members of the linkage family. This study demonstrates that human disease genes can be identified using information provided by MitoP2.
SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online.
ACKNOWLEDGEMENTS
The MitoP2 project is funded by the German National Genome Network (German Ministry for Education and Research, grant 01GR0411), BFAM (Bioinformatics for the Functional Analysis of Mammalian Genomes) and the MitEURO consortium. Funding to pay the Open Access publication charges for this article was provided by the GSF Research Centre.
REFERENCES
Collins, F.S., Green, E.D., Guttmacher, A.E., Guyer, M.S. (2003) A vision for the future of genomics research Nature, 422, 835–847 .
Taylor, S.W., Fahy, E., Ghosh, S.S. (2003) Global organellar proteomics Trends Biotechnol, . 21, 82–88 .
Prokisch, H., Scharfe, C., Camp, D.G., II, Xiao, W., David, L., Andreoli, C., Monroe, M.E., Moore, R.J., Gritsenko, M.A., Kozany, C., et al. (2004) Integrative analysis of the mitochondrial proteome in yeast PLoS Biol, . 2, e160 .
Andreoli, C., Prokisch, H., Hortnagel, K., Mueller, J.C., Munsterkotter, M., Scharfe, C., Meitinger, T. (2004) MitoP2, an integrated database on mitochondrial proteins in yeast and man Nucleic Acids Res, . 32, D459–D462 .
Scharfe, C., Zaccaria, P., Hoertnagel, K., Jaksch, M., Klopstock, T., Dembowski, M., Lill, R., Prokisch, H., Gerbitz, K.D., Neupert, W., et al. (2000) MITOP, the mitochondrial proteome database: 2000 update Nucleic Acids Res, . 28, 155–158 .
Szklarz, L., Guiard, B., Rissler, M., Wiedemann, N., Kozjak, V., van der Laan, M., Lohaus, C., Marcus, K., Meyer, H., Chacinska, A., et al. (2005) Inactivation of the mitochondrial heat shock protein Zim17 leads to aggregation of matrix Hsp70s followed by plelotropic effects on morphology and protein biogenesis J. Mol. Biol, . 351, 206–218 .
van der Laan, M., Chacinska, A., Lind, M., Perschil, I., Sickmann, A., Meyer, H., Guiard, B., Meisinger, C., Pfanner, N., Rehling, P. (2005) Pam17 is required for architecture and translocation activity of the mitochondrial protein import motor Mol. Cell. Biol, . 25, 7449–7458 .
Elpeleg, O., Miller, C., Hershkovitz, E., Bitner-Glindzicz, M., Bondi-Rubinstein, G., Rahman, S., Pagnamenta, A., Eshhar, S., Saada, A. (2005) Deficiency of the ADP-forming succinyl-CoA synthase activity is associated with encephalomyopathy and mitochondrial DNA depletion Am. J. Hum. Genet, . 76, 1081–1086 .
Mootha, V.K., Lepage, P., Miller, K., Bunkenborg, J., Reich, M., Hjerrild, M., Delmonte, T., Villeneuve, A., Sladek, R., Xu, F., et al. (2003) Identification of a gene causing human cytochrome c oxidase deficiency by integrative genomics Proc. Natl Acad. Sci. USA, 100, 605–610 .
Tiranti, V., Hoertnagel, K., Carrozzo, R., Galimberti, C., Munaro, M., Granatiero, M., Zelante, L., Gasparini, P., Marzella, R., Rocchi, M., et al. (1998) Mutations of SURF-1 in Leigh disease associated with cytochrome c oxidase deficiency Am. J. Hum. Genet, . 63, 1609–1621 .
Reichert, A.S. and Neupert, W. (2004) Mitochondriomics or what makes us breathe Trends Genet, . 20, 555–562 .
Cherry, J.M., Ball, C., Weng, S., Juvik, G., Schmidt, R., Adler, C., Dunn, B., Dwight, S., Riles, L., Mortimer, R.K., et al. (1997) Genetic and physical maps of Saccharomyces cerevisiae Nature, 387, 67–73 .
Huh, W.K., Falvo, J.V., Gerke, L.C., Carroll, A.S., Howson, R.W., Weissman, J.S., O'Shea, E.K. (2003) Global analysis of protein localization in budding yeast Nature, 425, 686–691 .
Kumar, A., Cheung, K.H., Tosches, N., Masiar, P., Liu, Y., Miller, P., Snyder, M. (2002) The TRIPLES database: a community resource for yeast molecular biology Nucleic Acids Res, . 30, 73–75 .
Dimmer, K.S., Fritz, S., Fuchs, F., Messerschmitt, M., Weinbach, N., Neupert, W., Westermann, B. (2002) Genetic basis of mitochondrial function and morphology in Saccharomyces cerevisiae Mol. Biol. Cell, 13, 847–853 .
Steinmetz, L.M., Scharfe, C., Deutschbauer, A.M., Mokranjac, D., Herman, Z.S., Jones, T., Chu, A.M., Giaever, G., Prokisch, H., Oefner, P.J., et al. (2002) Systematic screen for human disease genes in yeast Nature Genet, . 31, 400–404 .
DeRisi, J.L., Iyer, V.R., Brown, P.O. (1997) Exploring the metabolic and genetic control of gene expression on a genomic scale Science, 278, 680–686 .
Lascaris, R., Bussemaker, H.J., Boorsma, A., Piper, M., van der Spek, H., Grivell, L., Blom, J. (2003) Hap4p overexpression in glucose-grown Saccharomyces cerevisiae induces cells to enter a novel metabolic state Genome Biol, . 4, R3 .
Pflieger, D., Le Caer, J.P., Lemaire, C., Bernard, B.A., Dujardin, G., Rossier, J. (2002) Systematic identification of mitochondrial proteins by LC-MS/MS Anal. Chem, . 74, 2400–2406 .
Sickmann, A., Reinders, J., Wagner, Y., Joppich, C., Zahedi, R., Meyer, H.E., Schonfisch, B., Perschil, I., Chacinska, A., Guiard, B., et al. (2003) The proteome of Saccharomyces cerevisiae mitochondria Proc. Natl Acad. Sci. USA, 100, 13207–13212 .
Ghaemmaghami, S., Huh, W.K., Bower, K., Howson, R.W., Belle, A., Dephoure, N., O'Shea, E.K., Weissman, J.S. (2003) Global analysis of protein expression in yeast Nature, 425, 737–741 .
von Mering, C., Krause, R., Snel, B., Cornell, M., Oliver, S.G., Fields, S., Bork, P. (2002) Comparative assessment of large-scale data sets of protein–protein interactions Nature, 417, 399–403 .
Claros, M.G. (1995) MitoProt, a Macintosh application for studying mitochondrial proteins Comput. Appl. Biosci, . 11, 441–447 .
Guda, C., Fahy, E., Subramaniam, S. (2004) MITOPRED: a genome-scale method for prediction of nucleus-encoded mitochondrial proteins Bioinformatics, 20, 1785–1794 .
Nakai, K. and Horton, P. (1999) PSORT: a program for detecting sorting signals in proteins and predicting their subcellular localization Trends Biochem. Sci, . 24, 34–36 .
Small, I., Peeters, N., Legeai, F., Lurin, C. (2004) Predotar: a tool for rapidly screening proteomes for N-terminal targeting sequences Proteomics, 4, 1581–1590 .
Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Miller, W., Lipman, D.J. (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res, . 25, 3389–3402 .
Boeckmann, B., Bairoch, A., Apweiler, R., Blatter, M.C., Estreicher, A., Gasteiger, E., Martin, M.J., Michoud, K., O'Donovan, C., Phan, I., et al. (2003) The Swiss-Prot protein knowledgebase and its supplement TrEMBL in 2003 Nucleic Acids Res, . 31, 365–370 .
Ruepp, A., Zollner, A., Maier, D., Albermann, K., Hani, J., Mokrejs, M., Tetko, I., Guldener, U., Mannhaupt, G., Munsterkotter, M., et al. (2004) The FunCat, a functional annotation scheme for systematic classification of proteins from whole genomes Nucleic Acids Res, . 32, 5539–5545 .
Bult, C.J., Blake, J.A., Richardson, J.E., Kadin, J.A., Eppig, J.T., Baldarelli, R.M., Barsanti, K., Baya, M., Beal, J.S., Boddy, W.J., et al. (2004) The Mouse Genome Database (MGD): integrating biology with the genome Nucleic Acids Res, . 32, D476–D481 .
Gaucher, S.P., Taylor, S.W., Fahy, E., Zhang, B., Warnock, D.E., Ghosh, S.S., Gibson, B.W. (2004) Expanded coverage of the human heart mitochondrial proteome using multidimensional liquid chromatography coupled with tandem mass spectrometry J. Proteome. Res, . 3, 495–505 .
Taylor, S.W., Fahy, E., Zhang, B., Glenn, G.M., Warnock, D.E., Wiley, S., Murphy, A.N., Gaucher, S.P., Capaldi, R.A., Gibson, B.W., et al. (2003) Characterization of the human heart mitochondrial proteome Nat. Biotechnol, . 21, 281–286 .
Da Cruz, S., Xenarios, I., Langridge, J., Vilbois, F., Parone, P.A., Martinou, J.C. (2003) Proteomic analysis of the mouse liver mitochondrial inner membrane J. Biol. Chem, . 278, 41566–41571 .
Mootha, V.K., Bunkenborg, J., Olsen, J.V., Hjerrild, M., Wisniewski, J.R., Stahl, E., Bolouri, M.S., Ray, H.N., Sihag, S., Kamal, M., et al. (2003) Integrated analysis of protein composition, tissue diversity, and gene regulation in mouse mitochondria Cell, 115, 629–640 .
Ozawa, T., Sako, Y., Sato, M., Kitamura, T., Umezawa, Y. (2003) A genetic approach to identifying mitochondrial proteins Nat. Biotechnol, . 21, 287–293 .
Hamosh, A., Scott, A.F., Amberger, J., Bocchini, C., Valle, D., McKusick, V.A. (2002) Online Mendelian Inheritance in Man (OMIM), a knowledgebase of human genes and genetic disorders Nucleic Acids Res, . 30, 52–55 .
Drawid, A. and Gerstein, M. (2000) A Bayesian system integrating expression data with sequence patterns for localizing proteins: comprehensive application to the yeast genome J. Mol. Biol, . 301, 1059–1075 .
Marc, P., Margeot, A., Devaux, F., Blugeon, C., Corral-Debrinski, M., Jacq, C. (2002) Genome-wide analysis of mRNAs targeted to yeast mitochondria EMBO Rep, . 3, 159–164 .
Ohlmeier, S., Kastaniotis, A.J., Hiltunen, J.K., Bergmann, U. (2004) The yeast mitochondrial proteome, a study of fermentative and respiratory growth J. Biol. Chem, . 279, 3956–3979 .(H. Prokisch1,2,*, C. Andreoli1, U. Ahtin)
*To whom correspondence should be addressed. Tel: +49 89 3187 2890; Fax: +49 89 3187 3297; Email: prokisch@gsf.de
ABSTRACT
The MitoP2 database (http://www.mitop.de) integrates information on mitochondrial proteins, their molecular functions and associated diseases. The central database features are manually annotated reference proteins localized or functionally associated with mitochondria supplied for yeast, human and mouse. MitoP2 enables (i) the identification of putative orthologous proteins between these species to study evolutionarily conserved functions and pathways; (ii) the integration of data from systematic genome-wide studies such as proteomics and deletion phenotype screening; (iii) the prediction of novel mitochondrial proteins using data integration and the assignment of evidence scores; and (iv) systematic searches that aim to find the genes that underlie common and rare mitochondrial diseases. The data and analysis files are referenced to data sources in PubMed and other online databases and can be easily downloaded. MitoP2 users can explore the relationship between mitochondrial dysfunctions and disease and utilize this information to conduct systems biology approaches on mitochondria.
INTRODUCTION
The application of genomics to biology and medicine requires an understanding how specific gene variants contribute to phenotypes, in combination with a comprehensive knowledge of the ‘parts list’ of a cellular system and how these components are assembled into functional units (1). Mitochondria are ubiquitous and defined substructures of nucleated cells and lend themselves to systems biology approaches. However, in generic databases the annotation of mitochondrial proteins is often incomplete and does not always distinguish between proteins which have a confirmed mitochondrial subcellular localization and those which are only candidates according to preliminary experimental results or in silico predictions. For the human species, about half of the estimated 1500 proteins localized or functionally associated with mitochondria are known (2). Since the mitochondrial organelle is an evolutionarily conserved entity, systematic studies in model organisms are powerful to identify mitochondrial proteins in other organisms (3).
The MitoP2 database was created to consolidate and structure public information on mitochondrial proteins, their functions and associated human diseases (4,5). MitoP2 provides a wide variety of search functions to explore and download information and to access references in PubMed and other public databases. We have further expanded the manually annotated reference sets of mitochondrial proteins in yeast (522 proteins) and human (624 proteins), and have now added the section MitoP2-Mouse (615 proteins). For these three species, we integrated data from genome-wide approaches applied to the study of mitochondria, and assigned an evidence score of a candidate protein being mitochondrial (3). With the help of MitoP2, proteins involved in mitochondrial biogenesis and function have been identified and characterized (6,7). In addition, MitoP2 has enabled the identification of disease genes using positional candidate approaches (8–10).
MitoP2-YEAST
A wealth of information has been collected over the past several years from single gene and genome-wide studies of Saccharomyces cerevisiae (11). The list of yeast ORFs and protein annotations in MitoP2-Yeast are based on information in the Saccharomyces Genome Database (SGD; http://www.yeastgenome.org) (12). This MitoP2-Yeast update now provides annotated information for 522 mitochondrial reference proteins, which are based on experimental validation of each of these proteins. Recently, systematic cellular sublocalization studies estimated a total of 800 mitochondrial proteins presenting 12% of the currently known yeast genes (13,14). Therefore, 250–300 mitochondrial proteins are still missing. In order to identify these missing genes, we have validated and integrated genome-wide approaches applied to the study of mitochondria (3). MitoP2-Yeast datasets in Table 1 show 20 systematic approaches used for this purpose: phenotypes of single gene deletion mutant phenotypes (15,16); systematic subcellular localization studies (13,14); transcriptome datasets of differentially expressed genes including fermentable and non-fermentable growth conditions, the response to diauxic shift, and Hap4 transcription factor screening (3,17,18); proteome analyses of purified mitochondrial organelles (3,19,20); protein abundance measurements (21); and data from protein–protein interaction studies that include interactions to mitochondrial proteins (22). In addition to experimental datasets, mitochondrial proteins can be predicted in silico based on the presence of mitochondrial targeting sequences (23–26), and by sequence similarity to a known mitochondrial protein from other species (defined as bidirectional best BLAST hit or best BLAST hit with a score <1 x 10–10) (27). Data from each of these systematic studies can be searched and downloaded.
Table 1 Comparison of specificity and sensitivity for various approaches integrated in MitoP2 in determining the mitochondrial localization of proteins
Using the MitoP2-Yeast reference proteins, it is possible to analyze the specificity and sensitivity of the data from genome-wide studies (Figure 1). Specificity is defined as the proportion of proteins of a dataset which are part of the reference set, while sensitivity is the proportion of reference set proteins which is covered by the dataset. In order to identify putative mitochondrial proteins, we calculated a MitoP2 score for each protein, reflecting the specificity of combined approaches which identified the particular protein (3). To further improve these predictions, we used a new approach utilizing a support vector machine (SVM, http://svmlight.joachims.org). The SVMs are learning machines based on statistical learning theory used for solving classification tasks. We trained the SVM using the MitoP2-Yeast reference set (522 proteins) and a set of 519 proteins with a known localization to other cellular compartments collected from SGD (http://www.yeastgenome.org/). For each of the 1041 proteins, we defined a 20-dimensional vector using the datasets of 20 systematic studies (see Table 1). This resulted in a 20-dimensional input matrix, which was used to train the SVM (see also Supplementary Figure S1). After training, the SVM predicts mitochondrial proteins with a specificity of 78% and a sensitivity of 80% (SVM score >1). This analysis shows that a combination of datasets from genome-wide studies significantly increases the power of predicting mitochondrial proteins beyond the level achieved by any single study (Figure 1).
Figure 1 Systematic approaches to identify mitochondrial proteins. The yeast datasets were benchmarked against the mitochondrial reference set. Each point represents a dataset whose position is determined by benchmarking against the 522 reference proteins from MitoP2-Yeast. The different groups of approaches are highlighted using distinct colours: the bioinformatics datasets (purple) are PSORT (25), MitoProt >90 (23), Bayesian prediction (37), Predotar (26) and yeast proteins with human mitochondrial orthologs (MitoP2 database); the experimental datasets (blue) are as follows: hap4 expression (18), respiration induced expression (3), mitochondria localized ribosomes (38), deletion phenotype screen (16), tag localization (14), GFP localization (13), pet phenotypes (15), four mass spectrometry proteome studies (3,19,20,39) and high and medium confidence protein–protein interactions (PPI) (22) defined by interactions with known mitochondrial proteins (MitoP2 database). The predictive score for a mitochondrial protein (MitoP2 score; green) was based on the combination of the systematic datasets, calculated for different thresholds. The predictions using the SVM algorithm are shown in red for different thresholds.
MitoP2-HUMAN AND MitoP2-MOUSE
We manually annotated mitochondrial reference proteins for human (624) and mouse (615) that now cover about half of the estimated mitochondrial proteins in these two species. These reference proteins present a subset of all the protein entries in the database: MitoP2-Human contains 36 504 proteins and MitoP2-Mouse contains 32 422 proteins. These datasets have been downloaded from the Swiss-Prot database (http://www.expasy.org/sprot/) (28). To identify putative orthologue proteins between human and mouse we calculated a bidirectional best BLAST hit or a best BLAST hit with score <1 x 10–10 between the two datasets. For each MitoP2 protein, we extracted descriptions, chromosomal positions, subcellular localization and literature references from Swiss-Prot. In addition, functional annotations such as biological processes and functional categories were extracted from the Gene Ontology database (GO; http://www.geneontology.org/). For MitoP2-Mouse, we annotated functional descriptions according to the MIPS functional catalogue (29), and provided access to DNA and protein sequence information. Each of these protein annotations is accompanied by its PubMed reference link. Phenotypic information on available mouse models are provided by the Mouse Genome Informatics database (MGI; http://www.informatics.jax.org/) (30). To date, more than 50 mouse models carrying mutations or deletions of mitochondrial genes have been investigated. For researchers interested in studying these models, MitoP2 provides links to the International Gene Trap Consortium (IGTC; http://www.genetrap.org/) to access the related mouse cell lines.
MitoP2-Mouse and MitoP2-Human provide similar search options that allow single or combined searches for individual database components (Figure 2). Database searches and downloads can be performed using keywords, genes names and the selection of datasets from systematic studies. For MitoP2-Human, two proteome studies on mitochondrial organelles purified from heart tissue are available, which have been integrated under the ‘proteome’ category (31,32). For Mitop2-Mouse, we integrated the datasets from three high-throughput studies that include two proteome experiments (33,34) and a subcellular localization study using split-enhanced green fluorescent protein (EGFP) (35). For mouse proteins identified using these approaches, we identified the respective putative orthologue proteins in human, and vice versa. The number of proteins in human and mouse datasets differ in part due to missing proteins in either one of the species in Swiss-Prot. The MitoP2 category ‘transcriptome’ predicts gene relationships based on similarities of their expression profiles (34). In addition to sequence similarity searches between human, yeast and mouse, MitoP2 provides in silico predictions for mitochondrial proteins utilizing established algorithms such as MitoProt II (23), PSORT II (25), Predotar (26) and MITOPRED (24). These programs allow the prediction of subcellular localizations of proteins based on their amino acid sequences. To illustrate the different search functions, users can select PSORT II under MitoP2-Mouse to extract 4321 proteins that include 323 entries from the mitochondrial reference set (7%). Alternatively, one can perform combined searches, for example, by selecting PSORT II and a human proteome dataset ‘Hprot_01’ (31,32), which then generates a list of 176 proteins that includes 56% of the mitochondrial reference set. This comparison demonstrates the trade-off between sensitivity and specificity: the combination of datasets reduces the total number of proteins (sensitivity), while it increases the specificity for mitochondrial proteins.
Figure 2 Screenshot of the MitoP2-Mouse query page. The MitoP2 query page is structured according to various groups of search parameters provided by the database. The search options are either linked to the online references or an explanation for this selection is provided.
Each entry in MitoP2-Human and MitoP2-Mouse corresponds to a Swiss-Prot identifier with protein descriptions, annotated subcellular localization and sequence map positions according to UCSC genome browser (http://genome.ucsc.edu/). In addition, the single protein entry summarizes the information from in silico predictions, high-throughput experiments, the availability of mouse gene trap clones and the predictive MitoP2 score. An example for a single protein entry in MitoP2-Mouse, the adenine nucleotide (ADP/ATP) translocator 2, is shown in Figure 3. This figure shows in a matrix lane, the information available for this protein extracted from Swiss-Prot and the integrated genome-wide approaches, a list of functional annotations compiled from the MIPS catalogue that are linked to the Mouse Functional Genome Database (http://mips.gsf.de/genre/proj/mfungd/), PubMed links to the references, and a list of similar sequences from other species. The other parts of this entry, which are not shown in this figure, include a phenotype description of the associated mouse mutant, the Gene Ontology annotations for molecular protein functions and biological processes, literature references on protein functions and variants that are listed by author names and title and a table of Swiss-Prot references.
Figure 3 Example for protein entry in MitoP2-Mouse. As illustrated for the mitochondrial ADP/ATP carrier protein 2 (ADT2). MitoP2 provides for each protein entry the Swiss-Prot name and description, the chromosomal localization, results from mitochondrial prediction programs, data from proteome studies, available gene trap clones, functional annotations according to MIPS, PubMed reference links and homologous proteins in other species.
For genes implicated in a hereditary disease, MitoP2 provides a link to the corresponding entry in the Online Mendelian Inheritance in Man database (OMIM; http://www.ncbi.nlm.nih.gov) (36). To date, more than 120 of the 624 human mitochondrial proteins are known to be involved in a hereditary disease. Mitochondrial disorders have a diversity of debilitating phenotypes and include a wide variety of neurodegenerative processes, cardiovascular disorders, diabetes mellitus and several cancer types. Many of these disease genes function in the metabolism of amino acids, nucleic acid, fatty acids and lipids, and energy production. The MitoP2 database enables the systematic identification of candidate genes to study mitochondrial diseases (5). Elpeleg et al. (8), for example, mapped a locus for hereditary mtDNA depletions associated with mitochondrial encephalomyopathy to a 21 Mb interval on chromosome 13. The mapping coordinates (i.e. 13:40878920 and 13:61359487) were used as a selection criteria to prioritize MitoP2 candidate genes among the 113 genes predicted in this region. In combination with a MitoP2 score >60, three proteins were identified as disease candidate genes. One of these genes (SUCLA2), a mitochondrial reference protein identified in two proteome experiments, was found to be mutated in affected members of the linkage family. This study demonstrates that human disease genes can be identified using information provided by MitoP2.
SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online.
ACKNOWLEDGEMENTS
The MitoP2 project is funded by the German National Genome Network (German Ministry for Education and Research, grant 01GR0411), BFAM (Bioinformatics for the Functional Analysis of Mammalian Genomes) and the MitEURO consortium. Funding to pay the Open Access publication charges for this article was provided by the GSF Research Centre.
REFERENCES
Collins, F.S., Green, E.D., Guttmacher, A.E., Guyer, M.S. (2003) A vision for the future of genomics research Nature, 422, 835–847 .
Taylor, S.W., Fahy, E., Ghosh, S.S. (2003) Global organellar proteomics Trends Biotechnol, . 21, 82–88 .
Prokisch, H., Scharfe, C., Camp, D.G., II, Xiao, W., David, L., Andreoli, C., Monroe, M.E., Moore, R.J., Gritsenko, M.A., Kozany, C., et al. (2004) Integrative analysis of the mitochondrial proteome in yeast PLoS Biol, . 2, e160 .
Andreoli, C., Prokisch, H., Hortnagel, K., Mueller, J.C., Munsterkotter, M., Scharfe, C., Meitinger, T. (2004) MitoP2, an integrated database on mitochondrial proteins in yeast and man Nucleic Acids Res, . 32, D459–D462 .
Scharfe, C., Zaccaria, P., Hoertnagel, K., Jaksch, M., Klopstock, T., Dembowski, M., Lill, R., Prokisch, H., Gerbitz, K.D., Neupert, W., et al. (2000) MITOP, the mitochondrial proteome database: 2000 update Nucleic Acids Res, . 28, 155–158 .
Szklarz, L., Guiard, B., Rissler, M., Wiedemann, N., Kozjak, V., van der Laan, M., Lohaus, C., Marcus, K., Meyer, H., Chacinska, A., et al. (2005) Inactivation of the mitochondrial heat shock protein Zim17 leads to aggregation of matrix Hsp70s followed by plelotropic effects on morphology and protein biogenesis J. Mol. Biol, . 351, 206–218 .
van der Laan, M., Chacinska, A., Lind, M., Perschil, I., Sickmann, A., Meyer, H., Guiard, B., Meisinger, C., Pfanner, N., Rehling, P. (2005) Pam17 is required for architecture and translocation activity of the mitochondrial protein import motor Mol. Cell. Biol, . 25, 7449–7458 .
Elpeleg, O., Miller, C., Hershkovitz, E., Bitner-Glindzicz, M., Bondi-Rubinstein, G., Rahman, S., Pagnamenta, A., Eshhar, S., Saada, A. (2005) Deficiency of the ADP-forming succinyl-CoA synthase activity is associated with encephalomyopathy and mitochondrial DNA depletion Am. J. Hum. Genet, . 76, 1081–1086 .
Mootha, V.K., Lepage, P., Miller, K., Bunkenborg, J., Reich, M., Hjerrild, M., Delmonte, T., Villeneuve, A., Sladek, R., Xu, F., et al. (2003) Identification of a gene causing human cytochrome c oxidase deficiency by integrative genomics Proc. Natl Acad. Sci. USA, 100, 605–610 .
Tiranti, V., Hoertnagel, K., Carrozzo, R., Galimberti, C., Munaro, M., Granatiero, M., Zelante, L., Gasparini, P., Marzella, R., Rocchi, M., et al. (1998) Mutations of SURF-1 in Leigh disease associated with cytochrome c oxidase deficiency Am. J. Hum. Genet, . 63, 1609–1621 .
Reichert, A.S. and Neupert, W. (2004) Mitochondriomics or what makes us breathe Trends Genet, . 20, 555–562 .
Cherry, J.M., Ball, C., Weng, S., Juvik, G., Schmidt, R., Adler, C., Dunn, B., Dwight, S., Riles, L., Mortimer, R.K., et al. (1997) Genetic and physical maps of Saccharomyces cerevisiae Nature, 387, 67–73 .
Huh, W.K., Falvo, J.V., Gerke, L.C., Carroll, A.S., Howson, R.W., Weissman, J.S., O'Shea, E.K. (2003) Global analysis of protein localization in budding yeast Nature, 425, 686–691 .
Kumar, A., Cheung, K.H., Tosches, N., Masiar, P., Liu, Y., Miller, P., Snyder, M. (2002) The TRIPLES database: a community resource for yeast molecular biology Nucleic Acids Res, . 30, 73–75 .
Dimmer, K.S., Fritz, S., Fuchs, F., Messerschmitt, M., Weinbach, N., Neupert, W., Westermann, B. (2002) Genetic basis of mitochondrial function and morphology in Saccharomyces cerevisiae Mol. Biol. Cell, 13, 847–853 .
Steinmetz, L.M., Scharfe, C., Deutschbauer, A.M., Mokranjac, D., Herman, Z.S., Jones, T., Chu, A.M., Giaever, G., Prokisch, H., Oefner, P.J., et al. (2002) Systematic screen for human disease genes in yeast Nature Genet, . 31, 400–404 .
DeRisi, J.L., Iyer, V.R., Brown, P.O. (1997) Exploring the metabolic and genetic control of gene expression on a genomic scale Science, 278, 680–686 .
Lascaris, R., Bussemaker, H.J., Boorsma, A., Piper, M., van der Spek, H., Grivell, L., Blom, J. (2003) Hap4p overexpression in glucose-grown Saccharomyces cerevisiae induces cells to enter a novel metabolic state Genome Biol, . 4, R3 .
Pflieger, D., Le Caer, J.P., Lemaire, C., Bernard, B.A., Dujardin, G., Rossier, J. (2002) Systematic identification of mitochondrial proteins by LC-MS/MS Anal. Chem, . 74, 2400–2406 .
Sickmann, A., Reinders, J., Wagner, Y., Joppich, C., Zahedi, R., Meyer, H.E., Schonfisch, B., Perschil, I., Chacinska, A., Guiard, B., et al. (2003) The proteome of Saccharomyces cerevisiae mitochondria Proc. Natl Acad. Sci. USA, 100, 13207–13212 .
Ghaemmaghami, S., Huh, W.K., Bower, K., Howson, R.W., Belle, A., Dephoure, N., O'Shea, E.K., Weissman, J.S. (2003) Global analysis of protein expression in yeast Nature, 425, 737–741 .
von Mering, C., Krause, R., Snel, B., Cornell, M., Oliver, S.G., Fields, S., Bork, P. (2002) Comparative assessment of large-scale data sets of protein–protein interactions Nature, 417, 399–403 .
Claros, M.G. (1995) MitoProt, a Macintosh application for studying mitochondrial proteins Comput. Appl. Biosci, . 11, 441–447 .
Guda, C., Fahy, E., Subramaniam, S. (2004) MITOPRED: a genome-scale method for prediction of nucleus-encoded mitochondrial proteins Bioinformatics, 20, 1785–1794 .
Nakai, K. and Horton, P. (1999) PSORT: a program for detecting sorting signals in proteins and predicting their subcellular localization Trends Biochem. Sci, . 24, 34–36 .
Small, I., Peeters, N., Legeai, F., Lurin, C. (2004) Predotar: a tool for rapidly screening proteomes for N-terminal targeting sequences Proteomics, 4, 1581–1590 .
Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Miller, W., Lipman, D.J. (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res, . 25, 3389–3402 .
Boeckmann, B., Bairoch, A., Apweiler, R., Blatter, M.C., Estreicher, A., Gasteiger, E., Martin, M.J., Michoud, K., O'Donovan, C., Phan, I., et al. (2003) The Swiss-Prot protein knowledgebase and its supplement TrEMBL in 2003 Nucleic Acids Res, . 31, 365–370 .
Ruepp, A., Zollner, A., Maier, D., Albermann, K., Hani, J., Mokrejs, M., Tetko, I., Guldener, U., Mannhaupt, G., Munsterkotter, M., et al. (2004) The FunCat, a functional annotation scheme for systematic classification of proteins from whole genomes Nucleic Acids Res, . 32, 5539–5545 .
Bult, C.J., Blake, J.A., Richardson, J.E., Kadin, J.A., Eppig, J.T., Baldarelli, R.M., Barsanti, K., Baya, M., Beal, J.S., Boddy, W.J., et al. (2004) The Mouse Genome Database (MGD): integrating biology with the genome Nucleic Acids Res, . 32, D476–D481 .
Gaucher, S.P., Taylor, S.W., Fahy, E., Zhang, B., Warnock, D.E., Ghosh, S.S., Gibson, B.W. (2004) Expanded coverage of the human heart mitochondrial proteome using multidimensional liquid chromatography coupled with tandem mass spectrometry J. Proteome. Res, . 3, 495–505 .
Taylor, S.W., Fahy, E., Zhang, B., Glenn, G.M., Warnock, D.E., Wiley, S., Murphy, A.N., Gaucher, S.P., Capaldi, R.A., Gibson, B.W., et al. (2003) Characterization of the human heart mitochondrial proteome Nat. Biotechnol, . 21, 281–286 .
Da Cruz, S., Xenarios, I., Langridge, J., Vilbois, F., Parone, P.A., Martinou, J.C. (2003) Proteomic analysis of the mouse liver mitochondrial inner membrane J. Biol. Chem, . 278, 41566–41571 .
Mootha, V.K., Bunkenborg, J., Olsen, J.V., Hjerrild, M., Wisniewski, J.R., Stahl, E., Bolouri, M.S., Ray, H.N., Sihag, S., Kamal, M., et al. (2003) Integrated analysis of protein composition, tissue diversity, and gene regulation in mouse mitochondria Cell, 115, 629–640 .
Ozawa, T., Sako, Y., Sato, M., Kitamura, T., Umezawa, Y. (2003) A genetic approach to identifying mitochondrial proteins Nat. Biotechnol, . 21, 287–293 .
Hamosh, A., Scott, A.F., Amberger, J., Bocchini, C., Valle, D., McKusick, V.A. (2002) Online Mendelian Inheritance in Man (OMIM), a knowledgebase of human genes and genetic disorders Nucleic Acids Res, . 30, 52–55 .
Drawid, A. and Gerstein, M. (2000) A Bayesian system integrating expression data with sequence patterns for localizing proteins: comprehensive application to the yeast genome J. Mol. Biol, . 301, 1059–1075 .
Marc, P., Margeot, A., Devaux, F., Blugeon, C., Corral-Debrinski, M., Jacq, C. (2002) Genome-wide analysis of mRNAs targeted to yeast mitochondria EMBO Rep, . 3, 159–164 .
Ohlmeier, S., Kastaniotis, A.J., Hiltunen, J.K., Bergmann, U. (2004) The yeast mitochondrial proteome, a study of fermentative and respiratory growth J. Biol. Chem, . 279, 3956–3979 .(H. Prokisch1,2,*, C. Andreoli1, U. Ahtin)