Kentaro Tomii

MitoFates: Improved Prediction of Mitochondrial Targeting Sequences and their Cleavage Sites

Mitochondria provide numerous essential functions for cells and their dysfunction leads to a variety of diseases. Thus, obtaining a complete mitochondrial proteome should be a crucial step toward understanding the roles of mitochondria. Many mitochondrial proteins have been identified experimentally but a complete list is not yet available. To fill this gap, methods to computationally predict mitochondrial proteins from amino acid sequence have been developed and are widely used, but unfortunately, their accuracy is far from perfect. Here we describe MitoFates, an improved prediction method for cleavable N-terminal mitochondrial targeting signals (presequences) and their cleavage sites. MitoFates introduces novel sequence features including positively charged amphiphilicity, presequence motifs, and position weight matrices modeling the presequence cleavage sites. These features are combined with classical ones such as amino acid composition and physico-chemical properties as input to a standard support vector machine classifier. On independent test data, MitoFates attains better performance than existing predictors in both detection of presequences and in predicting their cleavage sites. We used MitoFates to look for undiscovered mitochondrial proteins from 42,217 human proteins (including isoforms such as alternative splicing or translation initiation variants). MitoFates predicts 1167 genes to have at least one isoform with a presequence. Five-hundred and eighty of these genes were not annotated as mitochondrial in either UniProt or Gene Ontology. Interestingly, these include candidate regulators of parkin translocation to damaged mitochondria, and also many genes with known disease mutations, suggesting that careful investigation of MitoFates predictions may be helpful in elucidating the role of mitochondria in health and disease. MitoFates is open source with a convenient web server publicly available.

Tam41 is a CDP-diacylglycerol synthase required for cardiolipin biosynthesis in mitochondria

CDP-diacylglycerol (CDP-DAG) is central to the phospholipid biosynthesis pathways in cells. A prevailing view is that only one CDP-DAG synthase named Cds1 is present in both the endoplasmic reticulum (ER) and mitochondrial inner membrane (IM) and mediates generation of CDP-DAG from phosphatidic acid (PA) and CTP. However, we demonstrate here by using yeast Saccharomyces cerevisiae as a model organism that Cds1 resides in the ER but not in mitochondria, and that Tam41, a highly conserved mitochondrial maintenance protein, directly catalyzes the formation of CDP-DAG from PA in the mitochondrial IM. We also find that inositol depletion by overexpressing an arrestin-related protein Art5 partially restores the defects of cell growth and CL synthesis in the absence of Tam41. The present findings unveil the missing step of the cardiolipin synthesis pathway in mitochondria as well as the flexibile regulation of phospholipid biosynthesis to respond to compromised CDP-DAG synthesis in mitochondria.

Predicting mostly disordered proteins by using structure-unknown protein data

Predicting intrinsically disordered proteins is important in structural biology because they are thought to carry out various cellular functions even though they have no stable three-dimensional structure. We know the structures of far more ordered proteins than disordered proteins. The structural distribution of proteins in nature can therefore be inferred to differ from that of proteins whose structures have been determined experimentally. We know many more protein sequences than we do protein structures, and many of the known sequences can be expected to be those of disordered proteins. Thus it would be efficient to use the information of structure-unknown proteins in order to avoid training data sparseness. We propose a novel method for predicting which proteins are mostly disordered by using spectral graph transducer and training with a huge amount of structure-unknown sequences as well as structure-known sequences. When the proposed method was evaluated on data that included 82 disordered proteins and 526 ordered proteins, its sensitivity was 0.723 and its specificity was 0.977. It resulted in a Matthews correlation coefficient 0.202 points higher than that obtained using FoldIndex, 0.221 points higher than that obtained using the method based on plotting hydrophobicity against the number of contacts and 0.07 points higher than that obtained using support vector machines (SVMs). To examine robustness against training data sparseness, we investigated the correlation between two results obtained when the method was trained on different datasets and tested on the same dataset. The correlation coefficient for the proposed method is 0.14 higher than that for the method using SVMs. When the proposed SGT-based method was compared with four per-residue predictors (VL3, GlobPlot, DISOPRED2 and IUPred (long)), its sensitivity was 0.834 for disordered proteins, which is 0.052–0.523 higher than that of the per-residue predictors, and its specificity was 0.991 for ordered proteins, which is 0.036–0.153 higher than that of the per-residue predictors. The proposed method was also evaluated on data that included 417 partially disordered proteins. It predicted the frequency of disordered proteins to be 1.95% for the proteins with 5%–10% disordered sequences, 1.46% for the proteins with 10%–20% disordered sequences and 16.57% for proteins with 20%–40% disordered sequences. The proposed method, which utilizes the information of structure-unknown data, predicts disordered proteins more accurately than other methods and is less affected by training data sparseness.BackgroundPredicting intrinsically disordered proteins is important in structural biology because they are thought to carry out various cellular functions even though they have no stable three-dimensional structure. We know the structures of far more ordered proteins than disordered proteins. The structural distribution of proteins in nature can therefore be inferred to differ from that of proteins whose structures have been determined experimentally. We know many more protein sequences than we do protein structures, and many of the known sequences can be expected to be those of disordered proteins. Thus it would be efficient to use the information of structure-unknown proteins in order to avoid training data sparseness. We propose a novel method for predicting which proteins are mostly disordered by using spectral graph transducer and training with a huge amount of structure-unknown sequences as well as structure-known sequences.ResultsWhen the proposed method was evaluated on data that included 82 disordered proteins and 526 ordered proteins, its sensitivity was 0.723 and its specificity was 0.977. It resulted in a Matthews correlation coefficient 0.202 points higher than that obtained using FoldIndex, 0.221 points higher than that obtained using the method based on plotting hydrophobicity against the number of contacts and 0.07 points higher than that obtained using support vector machines (SVMs). To examine robustness against training data sparseness, we investigated the correlation between two results obtained when the method was trained on different datasets and tested on the same dataset. The correlation coefficient for the proposed method is 0.14 higher than that for the method using SVMs. When the proposed SGT-based method was compared with four per-residue predictors (VL3, GlobPlot, DISOPRED2 and IUPred (long)), its sensitivity was 0.834 for disordered proteins, which is 0.052–0.523 higher than that of the per-residue predictors, and its specificity was 0.991 for ordered proteins, which is 0.036–0.153 higher than that of the per-residue predictors. The proposed method was also evaluated on data that included 417 partially disordered proteins. It predicted the frequency of disordered proteins to be 1.95% for the proteins with 5%–10% disordered sequences, 1.46% for the proteins with 10%–20% disordered sequences and 16.57% for proteins with 20%–40% disordered sequences.ConclusionThe proposed method, which utilizes the information of structure-unknown data, predicts disordered proteins more accurately than other methods and is less affected by training data sparseness.

Structure of the N-terminal Domain of PEX1 AAA-ATPase: CHARACTERIZATION OF A PUTATIVE ADAPTOR-BINDING DOMAIN

Peroxisomes are responsible for several pathways in primary metabolism, including β-oxidation and lipid biosynthesis. PEX1 and PEX6 are hexameric AAA-type ATPases, both of which are indispensable in targeting over 50 peroxisomal resident proteins from the cytosol to the peroxisomes. Although the tandem AAA-ATPase domains in the central region of PEX1 and PEX6 are highly similar, the N-terminal sequences are unique. To better understand the distinct molecular function of these two proteins, we analyzed the unique N-terminal domain (NTD) of PEX1. Extensive computational analysis revealed weak similarity (<10% identity) of PEX1 NTD to the N-terminal domains of other membrane-related type II AAA-ATPases, such as VCP (p97) and NSF. We have determined the crystal structure of mouse PEX1 NTD at 2.05-Å resolution, which clearly demonstrated that the domain belongs to the double-ψ-barrel fold family found in the other AAA-ATPases. The N-domains of both VCP and NSF are structural neighbors of PEX1 NTD with a 2.7- and 2.1-Å root mean square deviation of backbone atoms, respectively. Our findings suggest that the supradomain architecture, which is composed of a single N-terminal domain followed by tandem AAA domains, is a common feature of organellar membrane-associating AAA-ATPases. We propose that PEX1 functions as a protein unfoldase in peroxisomal biogenesis, using its N-terminal putative adaptor-binding domain.

Application of the MAFFT sequence alignment program to large data – reexamination of the usefulness of chained guide trees

Motivation: Large multiple sequence alignments (MSAs), consisting of thousands of sequences, are becoming more and more common, due to advances in sequencing technologies. The MAFFT MSA program has several options for building large MSAs, but their performances have not been sufficiently assessed yet, because realistic benchmarking of large MSAs has been difficult. Recently, such assessments have been made possible through the HomFam and ContTest benchmark protein datasets. Along with the development of these datasets, an interesting theory was proposed: chained guide trees increase the accuracy of MSAs of structurally conserved regions. This theory challenges the basis of progressive alignment methods and needs to be examined by being compared with other known methods including computationally intensive ones. Results: We used HomFam, ContTest and OXFam (an extended version of OXBench) to evaluate several methods enabled in MAFFT: (1) a progressive method with approximate guide trees, (2) a progressive method with chained guide trees, (3) a combination of an iterative refinement method and a progressive method and (4) a less approximate progressive method that uses a rigorous guide tree and consistency score. Other programs, Clustal Omega and UPP, available for large MSAs, were also included into the comparison. The effect of method 2 (chained guide trees) was positive in ContTest but negative in HomFam and OXFam. Methods 3 and 4 increased the benchmark scores more consistently than method 2 for the three datasets, suggesting that they are safer to use. Availability and Implementation: http://mafft.cbrc.jp/alignment/software/ Contact: katoh@ifrec.osaka-u.ac.jp Supplementary information: Supplementary data are available at Bioinformatics online.

Molecular architecture of the active mitochondrial protein gate

Dissecting the mitochondrial entry portal Mitochondria, the powerhouses of the cell, are mainly composed of proteins made in the cytosol. These newly synthesized proteins need to be imported across the organelles membrane through dedicated protein import machinery. Shiota et al. have worked out the architecture and mechanism of the mitochondrial protein import channel. Science, this issue p. 1544 A biochemical analysis reveals how the main protein entry gate of mitochondria imports preproteins. Mitochondria fulfill central functions in cellular energetics, metabolism, and signaling. The outer membrane translocator complex (the TOM complex) imports most mitochondrial proteins, but its architecture is unknown. Using a cross-linking approach, we mapped the active translocator down to single amino acid residues, revealing different transport paths for preproteins through the Tom40 channel. An N-terminal segment of Tom40 passes from the cytosol through the channel to recruit chaperones from the intermembrane space that guide the transfer of hydrophobic preproteins. The translocator contains three Tom40 β-barrel channels sandwiched between a central α-helical Tom22 receptor cluster and external regulatory Tom proteins. The preprotein-translocating trimeric complex exchanges with a dimeric isoform to assemble new TOM complexes. Dynamic coupling of α-helical receptors, β-barrel channels, and chaperones generates a versatile machinery that transports about 1000 different proteins.

Protein structure prediction using a variety of profile libraries and 3D verification

This study is intended to construct a useful method for fold recognition, regardless of whether the proteins to be compared are evolutionarily related. We developed several descendants of our profile–profile comparison method to make use of known structural information for protein structure prediction. Our prediction strategy in CASP6 is simple. For every CASP6 target, we derived target–template alignments from several different versions of profile–profile comparisons. We then constructed and exhaustively evaluated 3D models based on those alignments. Subsequently, we selected proper model(s) among them. We specifically addressed the validation of our simple approach for protein structure prediction through CASP6 because the fold recognition results of CASP5 revealed areas of improvement in the selection of good models. Consequently, we applied a more stringent method for 3D model evaluation this time. All generated models were evaluated based on a structural quality score calculated by both Verify3D and Prosa2003 programs. It turns out that the prediction results of our human group were supported by the results of three servers. The pipeline that we constructed for our human group prediction and human intervention were also greatly effective in improving prediction models, but the efficacy of our scheme for 3D model evaluation was obscure. Proteins 2005;Suppl 7:114–121.

Mammalian NUMT insertion is non-random

It is well known that remnants of partial or whole copies of mitochondrial DNA, known as Nuclear MiTochondrial sequences (NUMTs), are found in nuclear genomes. Since whole genome sequences have become available, many bioinformatics studies have identified putative NUMTs and from those attempted to infer the factors involved in NUMT creation. These studies conclude that NUMTs represent randomly chosen regions of the mitochondrial genome. There is less consensus regarding the nuclear insertion sites of NUMTs — previous studies have discussed the possible role of retrotransposons, but some recent ones have reported no correlation or even anti-correlation between NUMT sites and retrotransposons. These studies have generally defined NUMT sites using BLAST with default parameters. We analyze a redefined set of human NUMTs, computed with a carefully considered protocol. We discover that the inferred insertion points of NUMTs have a strong tendency to have high-predicted DNA curvature, occur in experimentally defined open chromatin regions and often occur immediately adjacent to A + T oligomers. We also show clear evidence that their flanking regions are indeed rich in retrotransposons. Finally we show that parts of the mitochondrial genome D-loop are under-represented as a source of NUMTs in primate evolution.

Novel Mechanism of Interaction of p85 Subunit of Phosphatidylinositol 3-Kinase and ErbB3 Receptor-derived Phosphotyrosyl Peptides

Ligand-activated and tyrosine-phosphorylated ErbB3 receptor binds to the SH2 domain of the p85 subunit of phosphatidylinositol 3-kinase and initiates intracellular signaling. Here, we studied the interactions between the N- (N-SH2) and C- (C-SH2) terminal SH2 domains of the p85 subunit of the phosphatidylinositol 3-kinase and eight ErbB3 receptor-derived phosphotyrosyl peptides (P-peptides) by using molecular dynamics, free energy, and surface plasmon resonance (SPR) analyses. In SPR analysis, these P-peptides showed no binding to the C-SH2 domain, but P-peptides containing a phospho-YXXM or a non-phospho-YXXM motif did bind to the N-SH2 domain. The N-SH2 domain has two phosphotyrosine binding sites in its N- (N1) and C- (N2) terminal regions. Interestingly, we found that P-peptides of pY1180 and pY1241 favored to bind to the N2 site, although all other P-peptides showed favorable binding to the N1 site. Remarkably, two phosphotyrosines, pY1178 and pY1243, which are just 63 amino acids apart from the pY1241 and pY1180, respectively, showed favorable binding to the N1 site. These findings indicate a possibility that the pair of phosphotyrosines, pY1178-pY1241 or pY1243-pY1180, will fold into an appropriate configuration for binding to the N1 and N2 sites simultaneously. Our model structures of the cytoplasmic C-terminal domain of ErbB3 receptor also strongly supported the speculation. The calculated binding free energies between the N-SH2 domain and P-peptides showed excellent qualitative agreement with SPR data with a correlation coefficient of 0.91. The total electrostatic solvation energy between the N-SH2 domain and P-peptide was the dominant factor for its binding affinity.

PoSSuM: a database of similar protein–ligand binding and putative pockets

Numerous potential ligand-binding sites are available today, along with hundreds of thousands of known binding sites observed in the PDB. Exhaustive similarity search for such vastly numerous binding site pairs is useful to predict protein functions and to enable rapid screening of target proteins for drug design. Existing databases of ligand-binding sites offer databases of limited scale. For example, SitesBase covers only ∼33 000 known binding sites. Inferring protein function and drug discovery purposes, however, demands a much more comprehensive database including known and putative-binding sites. Using a novel algorithm, we conducted a large-scale all-pairs similarity search for 1.8 million known and potential binding sites in the PDB, and discovered over 14 million similar pairs of binding sites. Here, we present the results as a relational database Pocket Similarity Search using Multiple-sketches (PoSSuM) including all the discovered pairs with annotations of various types. PoSSuM enables rapid exploration of similar binding sites among structures with different global folds as well as similar ones. Moreover, PoSSuM is useful for predicting the binding ligand for unbound structures, which provides important clues for characterizing protein structures with unclear functions. The PoSSuM database is freely available at http://possum.cbrc.jp/PoSSuM/.