Bioinformatics Advance Access originally published online on January 25, 2005
Bioinformatics 2005 21(9):1825-1830; doi:10.1093/bioinformatics/bti280
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Computational identification of human mitochondrial proteins based on homology to yeast mitochondrially targeted proteins
Metabolism Programme, Research Institute, The Hospital for Sick Children 555 University Avenue, Toronto, ON Canada, M5G1X8
*To whom correspondence should be addressed.
 |
Abstract |
|---|
 TOP Abstract INTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Motivation: Patients with defects of the mitochondrial respiratory chain due to mutations in nuclear genes are often undiagnosable due to the lack of information about the role of these genes. We therefore sought to produce a novel dataset of human nuclear-encoded mitochondrial proteins.
Results: We have used the web-based computer program Mitoprot to predict which proteins in the Saccharomyces cerevisiae genome are targeted to mitochondria. We then used this protein dataset to identify the homologous human proteins in the Unigene database using TBLASTN from NCBI. Human proteins with an Expectation value <10–5 and an Identity >30% were accepted as true homologues of the yeast proteins. These human proteins were then reanalyzed with Mitoprot. The final set of proteins comprises a dataset of 361 human mitochondrially targeted proteins with homology to all S.cerevisiae mitochondrially targeted proteins. One hundred twenty eight of these proteins are novel and are of unknown function.
Contact: bhr{at}sickkids.ca
Supplementary information: Supplementary tables will be available from http://www.sickkids.ca/Robinsonlab/
|
INTRODUCTION |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
The study of the proteins necessary for the functioning of the mitochondrion as an organelle, has in recent years become increasingly important given the central role of mitochondria in apoptosis and energy production. While ‘mitochondrial proteins’ can refer to those proteins encoded by the mitochondrial DNA, usually the term has a broader meaning, encompassing all the proteins present in the organelle, from proteins involved in mitochondrial replication and movement to those involved in respiration and ATP production.
Human mitochondrial DNA codes only for 13 protein components of the electron transport chain, in addition to a few ribosomal and transfer RNAs (Anderson et al., 1981). All the remaining mitochondrial genes are transcribed from the nuclear DNA. These genes are translated into proteins on cytoplasmic ribosomes and are then selectively targeted to the mitochondria. There they are translocated through the mitochondrial membranes and directed to their specific subcompartments within the organelle.
Most proteins destined for the mitochondrion possess an N-terminal leader sequence (MTS: mitochondrial targeting sequence) that is recognized by receptors in the outer mitochondrial membrane, facilitating import of the protein. The leader sequence is then proteolytically cleaved from the protein, revealing secondary signals that target the protein to the inner mitochondrial membrane, intermembrane space or matrix. The MTS has specific properties: its amino acid sequence usually forms an amphipathic 
-helix with one positive and one apolar face, followed by a hydrophobic -helix. The sequence is usually only 20–60 amino acids long, and is rich in hydrophobic, hydroxylated and basic residues, and low in acidic residues (Neupert, 1997). Not all proteins have an N-terminal MTS; some have an internal one instead. These include most of the members of the TOM and TIM complexes (translocase of the outer membrane and inner membrane proteins), the carrier proteins, such as the ATP/ADP translocase, and several subunits of the electron transport chain. The localization signal for these proteins lies within the mature protein and cannot be predicted.
Several computer programs exist which predict the subcellular localization of proteins, based on specific targeting sequences; these are summarized in Table 1. We chose to use Mitoprot for this study as it appeared to be the most robust (Claros and Vincens, 1996). This program analyzes an entire protein sequence, and predicts the probability of import into mitochondria, as well as the length of the MTS.
|
Mitochondrial databases
There are several online databases that contain a range of data pertaining to mitochondria. Some are restricted to collating human mitochondrial DNA polymorphisms and mutations and their relationship to disease (Kogelnik et al., 1998; Attimonelli et al., 2000), while others focus on the nuclear-encoded mitochondrial proteins. Mitop2 is an extensive database that contains both genetic and functional information on nuclear and mitochondrial DNA encoded proteins from Homo sapiens, Saccharomyces cerevisiae and Neurospora crassa (Andreoli et al., 2004). Other database collections of human mitochondrial proteins are Mitonuc (Attimonelli et al., 2002), MitoDat (Lemkin et al., 1996), Mitoproteome [based on both experimental and other database materials (Taylor et al., 2003b; Cotter et al., 2004)] and the Organelle Genome Database (GOBASE) (O'Brien et al., 2003). Most of these databases claim to contain a near complete listing of all mitochondrial proteins, and have usually been collated from a number of online database sources; however, there are extreme variations of the number and type of proteins in each database. A carefully curated compilation of mitochondrial gene products in yeast and animal cells has also been published, which lists the number of known yeast proteins to be 584 and human to be 734 (Schon, 2001).
Comparing computational to experimental methods in mitochondrial proteomics
Many groups have attempted to identify the total number of mitochondrial proteins, in humans and in other organisms. Several groups have used a proteomic approach to answer this problem experimentally. Large-scale proteomic analysis of the mitochondrion can identify many proteins, but no single method appears to be able to identify all the proteins from the mitochondrion of any organism. This is usually due to specific properties of certain proteins that make them difficult to isolate by just one method, or simply because the number of proteins present in the mitochondrion is unmanageable (reviewed in Jung et al., 2000; Lopez and Melov, 2002; Taylor et al., 2003a). There are also difficulties that can be ascribed to the dynamics of the mitochondrial proteome. Different tissues will express different complements of mitochondrial proteins, and many other factors, such as time, age of the individual, species and environmental effects, will cause differential gene expression. Thus no two proteomic studies will be able to isolate the same set of mitochondrial proteins.
Two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) has been the most common approach to identifying organellar proteins, but is inefficient for mitochondrial proteomes due to the presence of a large number of membrane-associated and alkaline proteins that cannot be resolved by this method. Small proteins are also difficult to resolve. Often proteins are represented by multiple spots due to post-translational modifications and/or modifications during sample preparation. The sheer number of mitochondrial proteins also means there will be problems with resolution. This method is usually biased towards cataloguing known proteins and not towards the identification of novel mitochondrial candidates. The largest scale proteomics study so far was carried out on S.cerevisiae by Kumar et al. (2002); 2744 yeast proteins were localized to subcellular compartments using high-throughput epitope tagging and immunofluorescent analysis. This is still a long way off from identifying every protein in S.cerevisiae (6178 proteins), but does allow the extrapolation of the probable number of proteins in the mitochondria (Kumar et al., 2002). Of the proteins, 13% are estimated to be mitochondrial, suggesting that there are 803 proteins. This study was also able to provide information about the sub-cellular localization of 1000 previously unknown proteins, including 100 mitochondrial proteins. However, it was later found to contain some anomalous results due to import failure of some of the tagged mitochondrial proteins. The mitochondrial proteome of S.cerevisiae has also been comprehensively analyzed using several separation methods and analysis by tandem mass spectrometry, from which 750 mitochondrial proteins were identified (Sickmann et al., 2003). Of these, 436 (58%) were known mitochondrial proteins, 208 have not yet been localized and 106 are known to be present in other subcellular compartments. Another group was able to identify 179 proteins from yeast mitochondria using SDS–PAGE gel separation and LC-MS/MS analysis; 19 of these are known to belong to other subcompartments, and 28 are of unknown localization (Pflieger et al., 2002).
The largest proteomics study of human mitochondrial proteins so far used sucrose gradient centrifugation, 1D-PAGE and protein identification by mass spectrometry to identify 615 proteins from the human heart (Taylor et al., 2003a). However, this approach was still not exhaustive as some known mitochondrial proteins were missed because they were either very hydrophobic or very small, and a few (19) non-mitochondrial proteins were identified as mitochondrial. Seventy-one unknown proteins were also identified.
While there has been definite progress in the experimental ability to isolate and characterize mitochondrial proteins over the last few years, these approaches are not able to isolate all the proteins in the mitochondria. Thus it is still difficult to define the mitochondrial proteome experimentally, and if there are possibilities of contamination or non-identification of proteins, then it will be impossible to catalogue all of the mitochondrial proteins. If it has proven so difficult to isolate all the mitochondrial proteins in an organism, then the chances of identifying novel proteins by these methods are limited.
There is only one study that has attempted to computationally identify and list all the mitochondrially targeted proteins in yeast. They propose 470 mitochondrial proteins to be present (out of 6905; 6.8% of the genome); however, their website-located database was not available for comparison (Huang and Li, 2004). Marcotte et al. (2000) were able to identify 361 nuclear-encoded, prokaryote-derived mitochondrial proteins using a novel method of phylogenetic profiling from the entire yeast genome (6217 ORFs). However, their dataset represents only 50% accuracy and 58% coverage, and does not include eukaryote-derived and organism-specific proteins and so will not be entirely comparable to ours. They extrapolate their numbers based on the proportion of conserved mitochondrial genes, to suggest there are 
630 mitochondrial genes in yeast. Using this method 31 entirely novel mitochondrial proteins were identified. Other computational methods (Table 1) merely extrapolate to the total number of mitochondrial proteins.
Compilation of a human mitochondrial protein dataset based on homology to yeast mitochondrial proteins
We have sought to compile a dataset of all the mitochondrial-targeted proteins in human, based on homology to S.cerevisiae mitochondrial proteins. This information will be invaluable for positional candidate genetic studies involving mitochondrial disease. Mitochondrial DNA (mtDNA) only encodes a few proteins and thus the disease profiles that accompany mtDNA mutations are fairly well defined. As mtDNA is maternally inherited, there is usually a distinct pedigree associated with these diseases as well as other specific phenomena due to heteroplasmy of mtDNA within the cells, and uneven distribution of mutated mtDNA in different cell types and between affected family members. It is often possible therefore, to identify those respiratory-deficient patients whose defects are due to mutations in nuclear DNA. However, these patients often do not have specific enough clinical phenotypes to identify the defective protein, and so this group remains undiagnosable. Clinical heterogeneity between patients and small families usually negates the possibility of using classical genetic techniques, such as linkage analysis to determine the causal gene defect. Biochemical assays can be equally ineffective in providing a clinical insight, as the reduction of many respiratory chain enzymes’ rates may not be caused by a mitochondrial defect. This is compounded by the fact that while the identity of a number of nuclear-encoded proteins expressed in the mitochondria are known, there are many that are still unknown.
We therefore sought to produce a novel dataset of nuclear-encoded proteins that are present in the mitochondria, by virtue of the fact that they possess an MTS, and are homologous to proteins in S.cerevisiae. We do not envision this to be a complete database of all mitochondrial proteins, but rather to contain only proteins that are highly homologous to the evolutionarily to distant yeast, and which will therefore have conserved roles in the mitochondria. That this database will provide invaluable information has been indicated by the identification of a large number of previously unknown proteins. Such a dataset would be extremely laborious to produce using the human protein database (which is still incomplete), and so, by first analyzing the entire yeast genome, those proteins that are most important in the mitochondria were identified. This dataset will improve our knowledge by highlighting the extent of conservation of yeast and human nuclear-encoded mitochondrial proteins as well as providing novel candidates for previously undiagnosable diseases. For known human proteins, the dataset can be searched by function, NCBI Locuslink number (now replaced by Entrez Gene) and keyword; for unknown proteins and cDNAs, searching by chromosome localization using the NCBI Map viewer will be invaluable.
|
METHODS |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Mitoprot software was available from ftp://ftp.ens.fr/pub/molbio/. The program is also available at the Mitop2 website: http://ihg.gsf.de/ihg/mitoprot.html
|
RESULTS AND DISCUSSION |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
The entire S.cerevisiae protein database, comprising 6178 proteins, was downloaded from the MIPS database on July 4, 2002 (Mewes et al., 2000). The proteins were analyzed with Mitoprot to determine their probability of import into the mitochondria. Only proteins with a probability of import >70% were accepted and these were divided into two groups: known proteins and unknown proteins (Supplementary Table 1). The known group comprised characterized and named proteins (440 proteins), while the unknown group comprised hypothetical proteins, questionable ORFs and those with ‘similarity to’ other proteins (531 proteins). The protein sequences were then used for TBLASTN (Altschul and Lipman, 1990) searches against the latest release of the human Unigene database to determine the most likely human orthologue for each yeast protein (Unigene release February 2003 and April 2003 v.160) (Boguski and Schuler, 1995). TBLASTN compares a protein query sequence against a nucleotide sequence database dynamically translated in all six reading frames and retrieves all matches that fall within the defined parameters. The top five TBLASTN results with an Expectation value <10–5 and an Identity value >30% were treated as significant matches between yeast and human proteins.
All of the significant TBLASTN results were individually scanned for redundancies (e.g. when one yeast protein matched with several human cDNAs obviously representing the same gene). Each of the final 1342 significant human cDNA matches were then physically mapped using Locuslink (Pruitt and Maglott, 2001) and the UCSC Human Genome Browser (April 2003 release) (Kent et al., 2002). These human proteins therefore represented a group of proteins that were highly homologous to those S.cerevisiae proteins that had a high probability of import into the mitochondria. To eliminate anomalies due to protein homologies among the species that were not representative of mitochondrial proteins (e.g. when a yeast protein identified matches with several human proteins from the same family), we subjected the 1342 human proteins to a second Mitoprot analysis. Again, only those proteins with >70% probability of import into the mitochondria were accepted. This group of human proteins comprised 475 proteins, and was further reduced to 361 proteins once all redundancies were removed (Supplementary Table 2).
The final group of proteins contains some unqualified proteins; this is intentional as some of the NCBI database proteins do not have enough information from which to determine their probability of import into the mitochondria. For some of the EST (expressed sequence tag) sequences in the Unigene database, a methionine amino acid that would indicate the start of translation cannot be identified from the protein sequence, and without this, a Mitoprot analysis cannot be carried out. There are also EST sequences in the database that have no associated protein sequence. These proteins have been kept in this dataset until subsequent annotation in GenBank is added for them, at which time their Mitoprot probability can be determined. The final dataset therefore intentionally overrepresents the number of computationally predicted human mitochondrial proteins.
We constructed this database to help identify novel human mitochondrial proteins that may be candidates for mitochondrial disease. While the diseases caused by mitochondrial DNA are few, there are many more diseases that can manifest from nuclear gene mutations. We sought to identify novel candidates for mitochondrial disease by utilizing the homology between humans and S.cerevisiae. The S.cerevisiae genome was fully sequenced in 1996, and comprises approximately 6000 genes, in comparison to the proposed 20 000–25 000 human protein-coding genes (International Human Genome Sequencing Consortium, 2004). S.cerevisiae thus provides a minimally complex eukaryotic model for humans. While there are many genes present in the human mitochondria that will not be present in the yeast, we constructed this database to help identify the most conserved proteins between the two organisms that have preserved their MTS, and thus represent a core of mitochondrial function. There are inherent differences between humans and S.cerevisiae that will affect the identification of some expected proteins in human mitochondria due to the presence of more recently evolved processes. In fact, the major limitation of this dataset is that there will be many mammalian proteins that do not have an orthologue in the evolutionarily distant yeast. For example, S.cerevisiae does not contain complex I, and has instead two alternative NADH dehydrogenases.
It has been predicted that 46% of identified human proteins have a yeast homologue (International Human Genome Sequencing Consortium, 2001), and so we believe this dataset will be a useful tool in identifying novel human mitochondrial proteins. Knowledge of yeast mitochondrial proteins has been invaluable in the past and has led to the identification of many human genes involved in disease. This is especially important for human genes that are identified by positional cloning, whereby the gene is first mapped to a chromosome region without reference to its function. By analysis of yeast homologues of human genes in the candidate region, it is often possible to ascribe a function to an unknown gene (reviewed in Foury and Kucej, 2002). The human genes in this dataset have been physically mapped, where possible, using information from Locuslink and the UCSC Genome Browser; thus the dataset can be searched for mitochondrial candidates based on positional cloning information. Thus it may be possible to identify a candidate for a mitochondrial disease that might not have been recognized before.
The most important use of this dataset will be in the identification of new candidates for mitochondrial diseases. By analyzing the protein dataset for the presence or absence of known mitochondrial proteins, we can determine the validity of the dataset. This can be demonstrated using the yeast protein YCR083W, which encodes TRX3, the mitochondrial thioredoxin protein. At the time of the initial TBLASTN searches, the yeast protein was identified as having ‘strong similarity to thioredoxin’ (MIPS database on July 4, 2002) and so is present in the ‘unknown proteins’ dataset. TBLASTN retrieved the top five human nucleotide matches, of which three were later removed due to redundancy. The two human protein sequences that were left were thioredoxin, TXN and thioredoxin 2, TXN2. Both were analyzed with Mitoprot and, at this point, TXN (which is cytosolic) was discarded as it did not have an MTS. Therefore, the true homologue of yeast mitochondrial TRX3 was identified as human mitochondrial TXN2.
The limitations of the dataset can be demonstrated using LRPPRC, a recently identified novel human protein involved in stabilizing the mRNA of COX1 (Mootha et al., 2003; Xu et al., 2004). Mutations in the gene result in a form of Leigh's disease, with complex IV deficiency, which is specific to the Saguenay-Lac Saint Jean region of the province of Quebec in Canada. The yeast homologue, PET309 (YLR067C), has a probability of import into the mitochondria of 79.7% as determined using Mitoprot. However, when using this protein in a TBLASTN search of the human Unigene database, no matches were found. Whereas the human protein LRPPRC has a Mitoprot score of 99.4%, the homology between the two proteins is too low (19%) to provide a match by these standards. Thus it is possible that the identity value of the species’ orthologues may be below the thresholds chosen, and hence some important proteins will be missed.
There are some problems with the parameters chosen for this study. Following the analysis of the yeast mitochondrial peroxiredoxin PRX1, it was shown that the yeast protein matched five human proteins with high similarity, but because only the top five matches were retrieved for the human protein analysis, only human PRDX1 and PRDX2 were identified, neither of which is mitochondrial. The human peroxiredoxin family comprises six members, and thus, PRDX3 was not identified as the true human homologue. For this same reason, redundancies also exist in the database, due to some instances of a single yeast protein matching several human proteins (a family), all having high Mitoprot probabilities. However, this information is still important for the human database, as the effectiveness of the database is based on the inclusion of as many human proteins that have high probability of mitochondrial import as possible.
The cellular distribution of each human proteins in the final dataset shows that the majority of the proteins (apart from those whose cellular distribution is ‘undetermined’) are indeed mitochondrial (Fig. 1). Of the proteins that have a known cellular localization, 62% are mitochondrial. Cellular localization was determined using information from Locuslink, GOBASE and the SGD database. If localization of the human protein could not be determined, the localization of the yeast protein was used (if known), but only if sufficient evidence existed that the proteins were truly homologous. Proteins that were localized based on yeast data are noted as such in Supplementary Table 2. If a protein has not been identified in the mitochondria, and has been identified in several other cellular localizations, only the primary one has been noted. It is possible that proteins assigned to other cellular compartments may also be present in the mitochondria, but have not been identified as yet. A lot of cellular localization assignments are computational, or inferred from incomplete experimental data. It is possible that some of these proteins may demonstrate localization to the mitochondria as well as other compartments, and this possibility cannot be discarded until there is further experimental proof.
|
There are some proteins that are definitely known not to be present in the mitochondria, and yet appear to have a valid MTS. These proteins usually have other targeting signals that override the apparent MTS. For example, there were a large number of histones identified, which are known to be localized in the nucleus, and yet have a computationally predicated MTS. However, when the amino acid sequence for histone 1, H3c (HIST1H3C) is analyzed using PSORT II, the protein is predicted to be nuclear, with targeting probabilities of 52.2% nuclear, 39.1% mitochondrial, 4.3% cytoplasmic and 4.3% endoplasmic reticulum.
Figure 2 shows the distribution of function of the mitochondrial human proteins: the largest group of proteins is of ‘unknown function’. The second largest group is of proteins that are known not to be mitochondrial (derived from Fig. 1) and thus their extramitochondrial function is not noted here. The other protein functions demonstrate a typical distribution of mitochondrial roles, such as oxidative phosphorylation, protein synthesis and tricarboxylic acid cycle function.
|
The recent increase in the number of published mitochondrial protein databases based on proteomic analysis projects has created competition for the development of complete organellar protein databases; however, these projects are fraught with experimental limitations. While the projects do provide some novel information with respect to previously unknown mitochondrial proteins, there is no guarantee that any such database is complete. Both experimental and computational methods have their limitations. We have strived to produce a complete database that contains all the evolutionarily conserved proteins between human and yeast that have a computationally predictable MTS. This database is different from others that have been produced in that the protein content is much smaller, but it provides information that could not be determined experimentally. There are 128 human proteins in this database that are completely novel; they have no known function and no known cellular localization. Yet all of these proteins have >70% probability of import into the mitochondria, and have high homology to similarly determined proteins in S.cerevisiae. These proteins will be invaluable for further studies to determine their validity as candidates for mitochondrial disease.
|
Acknowledgments |
|---|
We would like to thank the Centre for Computational Biology at The Hospital for Sick Children for help with the database searches. This work was supported by the Canadian Institute for Health Research and a Fellowship from the Jacob's Ladder Foundation (J.M.C).
|
Footnotes |
|---|
Present address: MRC Dunn Human Nutrition Unit, Wellcome Trust/MRC Building, Hills Road, Cambridge CB2 2XY, UK. 
Received on October 15, 2004; revised on December 14, 2004; accepted on January 17, 2005
|
REFERENCES |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Altschul, S.F. and Lipman, D.J. (1990) Protein database searches for multiple alignments. Proc. Natl Acad. Sci. USA, 87, 5509–5513
Anderson, S., Bankier, A.T., Barrell, B.G., de Bruijn, M.H., Coulson, A.R., Drouin, J., Eperon, I.C., Nierlich, D.P., Roe, B.A., Sanger, F., et al. (1981) Sequence and organization of the human mitochondrial genome. Nature, 290, 457–465[CrossRef][Medline].
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
Attimonelli, M., Altamura, N., Benne, R., Brennicke, A., Cooper, J.M., D'Elia, D., Montalvo, A., Pinto, B., De Robertis, M., Golik, P., et al. (2000) MitBASE: a comprehensive and integrated mitochondrial DNA database. The present status. Nucleic Acids Res., 28, 148–152
Attimonelli, M., Catalano, D., Gissi, C., Grillo, G., Licciulli, F., Liuni, S., Santamaria, M., Pesole, G., Saccone, C. (2002) MitoNuc: a database of nuclear genes coding for mitochondrial proteins. Update 2002. Nucleic Acids Res., 30, 172–173
Boguski, M.S. and Schuler, G.D. (1995) ESTablishing a human transcript map. Nat. Genet., 10, 369–371[CrossRef][Web of Science][Medline].
Claros, M.G. (1995) MitoProt, a Macintosh application for studying mitochondrial proteins. Comput. Appl. Biosci., 11, 441–447
Claros, M.G. and Vincens, P. (1996) Computational method to predict mitochondrially imported proteins and their targeting sequences. Eur. J. Biochem., 241, 779–786[Web of Science][Medline].
Cotter, D., Guda, P., Fahy, E., Subramaniam, S. (2004) MitoProteome: mitochondrial protein sequence database and annotation system. Nucleic Acids Res., 32, D463–D467
Emanuelsson, O., Nielsen, H., Brunak, S., von Heijne, G. (2000) Predicting subcellular localization of proteins based on their N-terminal amino acid sequence. J. Mol. Biol., 300, 1005–1016[CrossRef][Web of Science][Medline].
Foury, F. and Kucej, M. (2002) Yeast mitochondrial biogenesis: a model system for humans? Curr. Opin. Chem. Biol., 6, 106–111[CrossRef][Web of Science][Medline].
Huang, Y. and Li, Y. (2004) Prediction of protein subcellular locations using fuzzy k-NN method. Bioinformatics, 20, 21–28
International Human Genome Sequencing Consortium. (2001) Initial sequencing and analysis of the human genome. Nature, 409, 860–921[CrossRef][Medline].
International Human Genome Sequencing Consortium. (2004) Finishing the euchromatic sequence of the human genome. Nature, 431, 931–945[CrossRef][Medline].
Jung, E., Heller, M., Sanchez, J.C., Hochstrasser, D.F. (2000) Proteomics meets cell biology: the establishment of subcellular proteomes. Electrophoresis, 21, 3369–3377[CrossRef][Web of Science][Medline].
Kent, W.J., Sugnet, C.W., Furey, T.S., Roskin, K.M., Pringle, T.H., Zahler, A.M., Haussler, D. (2002) The human genome browser at UCSC. Genome Res., 12, 996–1006
Kogelnik, A.M., Lott, M.T., Brown, M.D., Navathe, S.B., Wallace, D.C. (1998) MITOMAP: a human mitochondrial genome database—1998 update. Nucleic Acids Res., 26, 112–115
Kumar, A., Agarwal, S., Heyman, J.A., Matson, S., Heidtman, M., Piccirillo, S., Umansky, L., Drawid, A., Jansen, R., Liu, Y., et al. (2002) Subcellular localization of the yeast proteome. Genes Dev., 16, 707–719
Lemkin, P.F., Chipperfield, M., Merril, C., Zullo, S. (1996) A World Wide Web (WWW) server database engine for an organelle database, MitoDat. Electrophoresis, 17, 566–572[CrossRef][Web of Science][Medline].
Lopez, M.F. and Melov, S. (2002) Applied proteomics: mitochondrial proteins and effect on function. Circ. Res., 90, 380–389
Marcotte, E.M., Xenarios, I., van Der Bliek, A.M., Eisenberg, D. (2000) Localizing Proteins in the cell from their Phylogenetic profiles. Proc. Natl Acad. Sci. USA, 97, 12115–12120
Mewes, H.W., Frishman, D., Gruber, C., Geier, B., Haase, D., Kaps, A., Lemcke, K., Mannhaupt, G., Pfeiffer, F., Schuller, C., Stocker, S., Weil, B. (2000) MIPS: a database for genomes and protein sequences. Nucleic Acids Res., 28, 37–40
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
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[CrossRef][Web of Science][Medline].
Neupert, W. (1997) Protein import into mitochondria. Annu. Rev. Biochem., 66, 863–917[CrossRef][Web of Science][Medline].
O'Brien, E.A., Badidi, E., Barbasiewicz, A., deSousa, C., Lang, B.F., Burger, G. (2003) GOBASE—a database of mitochondrial and chloroplast information. Nucleic Acids Res., 31, 176–178
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[Medline].
Pruitt, K.D. and Maglott, D.R. (2001) RefSeq and LocusLink: NCBI gene-centered resources. Nucleic Acids Res., 29, 137–140
Schon, E.A. (2001) Gene products present in mitochondria of yeast and animal cells. In Liza, A. and Pon, E.A.S. (Eds.). Methods in Cell Biology: Mitochondria, , San Diego, CA Academic Press 65, , pp. 463–482.
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
Taylor, S.W., Fahy, E., Ghosh, S.S. (2003a) Global organellar proteomics. Trends Biotechnol., 21, 82–88[CrossRef][Web of Science][Medline].
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., Ghosh, S.S. (2003b) Characterization of the human heart mitochondrial proteome. Nat. Biotechnol., 21, 281–286[CrossRef][Web of Science][Medline].
Xu, F., Morin, C., Mitchell, G., Ackerley, C., Robinson, B.H. (2004) The role of the LRPPRC gene in cytochrome oxidase assembly: mutation causes lowered levels of COX I and COX III mRNA. Biochem. J., 382, 331–336[CrossRef][Web of Science][Medline].
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Kumar, R. Verma, and G. P. S. Raghava Prediction of Mitochondrial Proteins Using Support Vector Machine and Hidden Markov Model J. Biol. Chem., March 3, 2006; 281(9): 5357 - 5363. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||