Hidden localization motifs: naturally occurring peroxisomal targeting signals in non-peroxisomal proteins
© Neuberger et al.; licensee BioMed Central Ltd. 2004
Received: 25 May 2004
Accepted: 9 November 2004
Published: 30 November 2004
Can sequence segments coding for subcellular targeting or for posttranslational modifications occur in proteins that are not substrates in either of these processes? Although considerable effort has been invested in achieving low false-positive prediction rates, even accurate sequence-analysis tools for the recognition of these motifs generate a small but noticeable number of protein hits that lack the appropriate biological context but cannot be rationalized as false positives.
We show that the carboxyl termini of a set of definitely non-peroxisomal proteins with predicted peroxisomal targeting signals interact with the peroxisomal matrix protein receptor peroxin 5 (PEX5) in a yeast two-hybrid test. Moreover, we show that examples of these proteins - chicken lysozyme, human tyrosinase and the yeast mitochondrial ribosomal protein L2 (encoded by MRP7) - are imported into peroxisomes in vivo if their original sorting signals are disguised. We also show that even prokaryotic proteins can contain peroxisomal targeting sequences.
Thus, functional localization signals can evolve in unrelated protein sequences as a result of neutral mutations, and subcellular targeting is hierarchically organized, with signal accessibility playing a decisive role. The occurrence of silent functional motifs in unrelated proteins is important for the development of sequence-based function prediction tools and the interpretation of their results. Silent functional signals have the potential to acquire importance in future evolutionary scenarios and in pathological conditions.
For an increasing number of otherwise uncharacterized protein sequences from genome-sequencing projects, function assignment is attempted solely with in silico prediction methods, as reliable and cost-effective large-scale experimental methods are not available. In addition to sequence homology and annotation transfer considerations , these function assignments increasingly rely on algorithms that recognize protein-sequence features responsible for posttranslational modifications, subcellular localization and interactions with specific domains of other proteins.
Although considerable effort has been invested in achieving low false-positive prediction rates, our experience with tools for recognizing glycosyl phosphatidylinositol (GPI) lipid [2, 3] and myristoyl [4–6] anchor attachment sites and for predicting potential targets for PTS1-dependent translocation to peroxisomes  shows that a small but noticeable number of proteins without appropriate biological context (for example with contradictory subcellular localization or in taxa without the modifying enzyme or receptor) are systematically hit by these tools. For example, we found more than a dozen metazoan lysozymes [7, 8], known extracellular proteins, that are predicted to have carboxyl termini with a functional peroxisomal targeting signal 1 (PTS1) region.
Are these false-positive predictions? All three of the sequence-analysis tools mentioned above check query sequences for a recognition pattern that is explicitly described in terms of its physical properties and it is possible to check the concordance between pattern descriptions and query sequence individually. Nevertheless, this visual inspection is frequently unable to rationalize the findings as false-positive predictions, as all known components of the pattern appear to be present. Even in the case of high accuracy of the prediction tool, an erroneous prediction cannot be excluded. Alternatively, these predicted sequence motifs may occur by chance and be functional in an appropriate test system, but still have no biological meaning because the necessary cellular context is absent in vivo. Only experimental tests can resolve this contradiction. As a case study, we report the results of an experimental analysis that demonstrates the existence of naturally occurring peroxisomal targeting signals in several known non-peroxisomal proteins. We also discuss the evolutionary perspective of functional localization signals in unrelated proteins as well as the consequences for experimental localization determination and function prediction from sequence.
The major mechanism for targeting proteins to the matrix of peroxisomes, which are membrane-bounded organelles  of eukaryotic cells, is initiated in the cytoplasm by interaction of the receptor protein peroxin 5 (PEX5) with the carboxy-terminal signal PTS1 on the target protein [10, 11]. This signal consists of three regions of sequence comprising approximately 12 residues [12, 13]. It is composed of the most carboxy-terminal tripeptide (classically, the -SKL terminus), preceded by a region of around four residues (which interact with the surface at the mouth of the PEX5 binding cavity), and a solvent-accessible (or easily unfoldable) stretch of around five residues further upstream. The PTS1-prediction program 'PTS1'  identifies PTS1 signals in query protein sequences by evaluating their carboxy-terminal ends with respect to features necessary for interaction with the tetratricopeptide repeats of PEX5. The predictor's scoring function searching for this motif within the 12 carboxy-terminal residues achieves an estimated sensitivity of 90% and a selectivity above 99% .
The carboxyl termini of several non-peroxisomal proteins interact with PEX5
Screening of SWISS-PROT  entries with the PTS1 predictor identified proteins from several families that are clearly not peroxisomal but score highly and are predicted as PEX5 targets [7, 8]. We were not able to rationalize these results as false predictions as the proteins' carboxyl termini did not deviate from the generalized PTS1 sequence pattern . To verify whether these proteins could indeed interact with PEX5, we tested the carboxyl termini of seven representative proteins in a yeast two-hybrid system: hen egg-white lysozyme (P00698, secreted); dog lysozyme C from milk (P81708); tyrosinase from human (P14679, a melanosomal type I membrane protein); frog tyrosinase (Q04604); Drosophila sevenless (P13368, a large transmembrane protein required for photoreceptor development); precursor of lysosomal bovine cathepsin D (P80209); and a mitochondrial ribosomal protein from yeast (P12687). We also examined the carboxyl terminus of a mouse dihydrofolate reductase construct with an added SKL peptide, which has been shown not to be imported into yeast peroxisomes [16, 17].
Results of the yeast-two hybrid interaction assays with PEX5
Activity† (Units/mg protein)
Activity† (Units/mg protein)
Mitochondrial ribosomal protein L2, or MRP7
The accessibility of the PTS1-like carboxyl terminus is critical
The fact that the peroxisomal translocation machinery fails to import naturally occurring mature proteins carrying PTS1 signals into peroxisomes in vivo could be explained by the non-accessibility of their carboxyl termini. These could either be hidden in the native structure of the mature protein or of its functional complexes, or competing translocation machineries could lead to a removal of the respective proteins from the cytosol before their recognition by PEX5.
The first possibility is exemplified by DHFR-SKL. The carboxy-terminal 16 residues of the DHFR-SKL construct (EKGIKYKFEVYEKSKL, sequences appended to DHFR are in bold type, see results in Table 1) interact with yeast PEX5 in the two-hybrid test but in vivo the complete construct is not imported into peroxisomes, thus confirming the prediction [16, 17]. For comparison, it should be noted that two other DHFR-derived constructs with slightly longer carboxyl termini (IKYKFEVYEKGGKSKL and IKYKFEVYEKKNIESKL) are predicted to be peroxisomally targeted. Their scores calculated with the PTS1 predictor  are 13.2 and 9.9, respectively (compare with data in Table 1). They were experimentally shown  to be translocated to peroxisomes. In the native three-dimensional structure of DHFR , the carboxyl terminus is part of a β-sheet that is buried in the fold, deprived of flexibility and accessibility. Seemingly, this structure prevents the carboxy-terminal appended residues SKL in the construct from entering the PEX5 binding cavity, whereas slightly longer carboxyl termini may do. In our two-hybrid test system, the carboxy-terminal 16-mers are always considered exposed as, in the non-native sequence environment of the carboxyl terminus of the GAL4 activation domain, they are free from interfering or blocking structural features. Thus, DHFR-SKL fails to be imported into peroxisomes because its carboxyl terminus is sequestered in the structure of the mature protein.
Competing targeting signals prevent translocation into peroxisomes despite the presence of PTS1-like carboxyl termini
Alternatively, functional PTS1 signals can be overruled by other localization signals . For instance, distribution of the mammalian alanine-glyoxylate amino transferase (AGT) between peroxisomes and mitochondria is regulated by the variable occurrence of an amino-terminal mitochondrial targeting signal in the mature protein (depending on the usage of two alternative transcription initiation sites) [19, 20].
Does a naturally occurring PTS1-like carboxyl terminus of a clearly non-peroxisomal protein that is capable of interacting with PEX5 indeed lead to in vivo import of the respective protein, provided that a potentially overruling sequence signal is eliminated? A set of three target proteins with amino-terminal leader sequences was chosen from Table 1. Chicken lysozyme (SWISS-PROT id P00698), a secreted enzyme, is one of the best characterized proteins and has an apparently accessible carboxyl terminus as deduced from its three-dimensional structure (Protein Data Bank (PDB) number 1H6M ). The corresponding carboxy-terminal 16-mer produces moderate β-galactosidase activity in the yeast two-hybrid assay (most of the other proteins in Table 1 appear to interact even more strongly with PEX5). Human tyrosinase (P14679) is a melanosomal marker protein that functions in the formation of pigments such as melanins. Yeast 60S ribosomal protein L2 (P12687), or MRP7, is a component of the large subunit of the mitochondrial ribosome.
Functional PTS1 sequences can occur in organisms without peroxisomes
The occurrence of silent PTS1s without a targeting role raises the question of whether such signals can also evolve in organisms that do not carry peroxisomes. To test this hypothesis, we extended Table 1 with a set of four predicted carboxyl termini from prokaryotic enzymes: Escherichia coli glutamate-1-semialdehyde 2,1-aminomutase (P23893), E. coli transaldolase A (P78258), Methanopyrus kandleri riboflavin synthase (NCBI-Refseq accession NP_613646) and Archaeoglobus fulgidus 2-nitropropane dioxygenase (NCBI-Refseq accession NP_070998). Indeed, these proteins harbor carboxyl termini that qualify as PTS1 signals (lower part of table 1). As confirmation, for the bacterial protein glutamate-1-semialdehyde 2,1-aminomutase (GSA) we used the same methodology for subcellular localization determination as for yeast MRP7. The resulting GFP-GSA construct is also imported into peroxisomes (Figure 2), demonstrating that its PTS1-like carboxyl terminus is functional in the mature protein.
In families of orthologous proteins, peroxisomal location and its targeting signal in the amino-acid sequence are not necessarily conserved. For example, in plants the five enzymes of the glyoxylate cycle are localized to peroxisomes, but in S. cerevisiae three of the five (aconitase, isocitrate lyase, and the respective malate dehydrogenase isoform) could not be found in peroxisomes . Thus, it is not surprising to find sporadically occurring PTS1 signals in protein families (see some examples in Table 1).
In dually localized proteins such as AGT , the PTS1 signal has a biological role as a targeting signal. However, the carboxyl termini of the proteins from Table 1 do not seem to fulfill any specific targeting function. We suggest that these PTS1 signals occur as a result of neutral mutation. The presence of a functional PTS1 signal would not lead to evolutionary pressure in this context because mislocalization is prevented by overriding the function of these sequences either by alternative exposure of amino-terminal signals or by steric carboxy-terminal inaccessibility.
We conclude that a PEX5-interacting sequence can evolve simply by mutational alterations in the carboxy-terminal region of a protein. Although shuffling of a carboxy-terminal exon cannot be excluded for other examples, the fact that the open reading frames (ORFs) of the carboxy-terminal exons for human tyrosinase (GenBank accession AP000720.4), fly sevenless (GenBank accession AE003484.2) and chicken lysozyme (GenBank accession AF410481.1) reach far into the functional domains of their proteins, rather supports an evolutionary mechanism of several point substitutions. The occurrence of functional PTS1 sequences in non-eukaryotic species further supports a stochastic model for the evolution of PEX5-interacting protein carboxyl termini.
In non-globular regions of proteins, sequences that code for targeting to other subcellular compartments, or for posttranslational modifications, might appear in similar ways during evolution. For example, the sequence motif coding for amino-terminal N-myristoylation of glycines behaves as an exchangeable functional module, as protein families do exist where it has been substituted by alternative sequence determinants that facilitate membrane association . This is exemplified by the Arabidopsis thaliana Rab5 ortholog Ara7 and its paralog Ara6. Ara7 is geranylgeranylated on carboxy-terminal cysteines just as Rab5 is in other species. However, the closely related paralog Ara6 lacks the carboxy-terminal cysteines and has an experimentally verified amino-terminal myristoylation motif .
Many of these signals seem to remain silent under normal physiological conditions (as is the case for the PTS1 signal in some metazoan lysozymes) but have the potential to become important in some future evolutionary scenarios or in pathological situations. Alternatively, the PTS1 signal might have become obsolete and the corresponding sequence segment is now subject to evolutionary alterations. Apparently, the cell exploits only a fraction of the potential molecular capabilities of its proteins.
Futhermore, subcellular targeting is organized in a hierarchy of cellular recognition mechanisms. The co-translational sorting into the ER serves as a first decision node. Posttranslational processes such as interaction with chaperones, folding, and covalent modifications are concomitant with the appropriate exposure of targeting signals. The amino-terminal signals are made first and are therefore favored when it comes to recognition by receptors. PEX5 needs only to categorize the remaining unsorted proteins with accessible carboxyl termini into 'stay here' or 'let's go into peroxisomes'. This might also explain why the PTS1 signal is comparatively short and permissive for a wide range of residues.
Clearly, the fact that functional sequences for subcellular targeting occur in unrelated proteins needs to be considered for prediction-tool development. The construction of a negative learning set (sequences without the specific localization signal) on the basis of proteins with differing cellular localization is problematic. For example, a set of non-peroxisomal but organellar localized , viral  or bacterial sequences might contain a considerable number of proteins that potentially interact with PEX5. Thus, such a set does not directly qualify for automated learning procedures or the assessment of false-positive prediction [27, 28].
Surprisingly, when Maurer-Stroh and Eisenhaber applied their myristoylation site predictor for eukaryotic proteins to bacterial proteomes , systematic hits were found despite the absence of known amino-terminal N-myristoyltransferases (NMT) in bacteria. Are these false-positive predictions? A literature search revealed that myristoylation by host NMTs has physiological relevance for several secreted proteins of intracellular bacterial parasites . Thus, the sequence motif coding for amino-terminal N-myristoylation is typical for eukaryotes but occurs also in bacteria. In many cases, it remains without phenotypic effect for bacteria but may become evolutionarily important in the case of host-parasite interactions.
In the case of the endothelin-converting enzyme 1 and the neprilysin-like zinc metallopeptidase family, the carboxy-terminal CXAW motif is a valid prenylation motif. This carboxy-terminus is functionally hidden because the protein is exported to the extracellular side of the cytomembrane and the carboxy-terminal residues are apparently involved in folding and enzyme function .
Clearly, the accessibility of the recognition motif in the substrate protein to the respective receptor or protein-modifying enzyme is a major issue. For PTS1 signal prediction from the amino-acid sequence, carboxy-terminal exposure needs to be assessed both from the steric point of view as well as in the context of competing translocation mechanisms. Analyzing only the carboxy-terminal dodecamer peptide [7, 13] might not suffice for reliable prediction of accessibility to the receptor, but a full solution would require sufficiently accurate three-dimensional structure prediction.
In databases, it should also be routine to flag proteins that contain several competing targeting signals with differing priority. Finally, silent localization signals might become active in mutant protein constructs and lead to non-native localizations, an issue that needs to be assessed especially in localization screens of proteins with uniformly incorporated fluorescent dyes such as GFP. It cannot be excluded that the subcellular location of a considerable number of proteins has not been correctly determined in published large-scale studies that rely on this methodology [30, 31].
To conclude, sequence segments coding for subcellular targeting or for posttranslational modifications can occur in proteins that are not substrates in either of these processes. Accurate prediction techniques reveal candidate proteins carrying hidden sequence signals. Many of these can be experimentally confirmed. In the case of the PTS1 predictor program, there is no reasonable argument to assume a difference in prediction accuracies for real and hidden PTS1s as, in both cases, productive interaction of the carboxyl terminus with PEX5 is the criterion for a functional PTS1.
Materials and methods
Oligonucleotides were purchased from MWG Biotech (Munich, Germany). The E. coli strain DH5α, Bethesda Research Laboratories) was used for all transformations and plasmid isolations. For the yeast two-hybrid-assay, the hybridized oligonucleotide pairs coded for the carboxy-terminal 16-mers of the selected proteins flanked by BamHI (5') and EcoRI (3') restriction sites. Each oligonucleotide pair was introduced into a BamHI-EcoRI-digested pGAD.GH fragment, generating plasmids containing the Gal4p activation domain in addition to the desired carboxy-terminal 16-mer extension (Gal4pAD-16mer). All pGAD.GH constructs were sequenced (VBC Genomics, Vienna, Austria). The plasmids pAH987 and hP87 contain the binding domain of Gal4p fused to the TPR domain of S. cerevisiae or Homo sapiens PEX5, respectively (Gal4pBD-TPR) .
Oligonucleotides used for the amplification of the GFP-constructs
DsRed2-SKL (COS7 cells)
DsRed-SKL (yeast cells)
Standard procedures were used for cloning of the GFP-MRP7 and GFP-GSA constructs including control sequencing (VBC Genomics). The plasmids expressing GFP and GFP-SKL under control of the MLS1 promoter were described previously . The DNA fragment coding for DsRed-SKL was obtained by PCR (for oligonucleotides, see Table 2; template pDsRed, Clontech) and cloned (BamHI-and partially with PstI) after the MLS1 promoter in the vector YEplac181. DNA fragments coding for MRP7 and GSA were obtained by PCR (see Table 2 for oligonucleotide sequences) and cloned (BamHI-SphI) in-frame with GFP to give rise to the expression of GFP-MRP7 and GFP-GSA, respectively, all of them under the control of the MLS1 promoter.
Yeast two-hybrid assay
According to the Matchmaker two-hybrid protocol, yeast strain PCY3 (MATα, his3Δ200, ade2-101, trp1Δ63, leu2, gal4Δ, gal80Δ, lys2::GAL1-HIS3, ura3::GAL1-lacZ)  was transformed with the Gal4pAD-16mer constructs (plasmid pGAD.GH) together with either pAH987 or hP87. Yeast transformants were selected and grown on minimal medium containing 2% glucose and supplemented with bases and amino acids as required (SC-leu-trp). For quantitative measurement of β-galactosidase activity in accordance with published techniques , yeast cells were grown in selective medium (SC-leu-trp) overnight at 30°C, diluted to A600 = 0.3 into the same medium and finally harvested at absorptions of A600 between 0.9 and 1.1.
In vivolocalization study in COS7 cells
COS7 cells were transfected with the pEGFP-C1-constructs and DsRed2-SKL by electroporation using 920 μF and 220 mV (Gene pulser II, Bio-Rad), grown on coverslips for 36 h, washed, fixed with 0.5% formaldehyde in PBS for 15 min and covered with geltol. Cells were analyzed using the Olympus BX51 fluorescence microscope (60 × enlargement).
In vivolocalization study in yeast cells
The yeast strain used in this study is S. cerevisiae CB80 (MATa, ura3-52, leu2-1, trp1-63, his3-200). Yeast transformants were selected and grown on minimum medium containing 0.67% yeast nitrogen bases without amino acids (Difco Laboratories), 2% glucose and amino acids (20-150 μg/ml) as required (SC-leu-ura). For fluorescence microscopy, yeast cells were grown at 30°C with shaking in selective media with 0.5% glucose as sole carbon source until the glucose concentration was very low (0.05%, usually 16 h), harvested by centrifugation and resuspended in the original volume of induction medium containing 0.67% yeast nitrogen bases without amino acids, 0.1% yeast extract, 30 mM potassium phosphate pH 6.0, 0.125% oleate, 0.2% Tween-80 and amino acids as required. Cells were grown for 16 h in induction medium and observed live for fluorescence. Briefly, cells were collected by centrifugation and washed twice in water. Cell pellets were resuspended in induction medium without oleate and aliquots were spotted onto multitest slides (ICN Biochemicals) previously coated with concanavalin A (6 mg/ml, Sigma). Cells were allowed to attach for 5 min at room temperature and the slides were washed twice with induction medium and a coverslip applied for observation. Fluorescence was viewed with a Zeiss Axioplan 2 fluorescence microscope using a 63 × (1.4 NA) lens. Digital images were captured with a Quantix CCD camera using Lightview software without further modification. The pictures were mounted and false-color overlays were made in Adobe Photoshop.
We wish to acknowledge the skilled technical assistance of Michael Schuster (Medical University, Vienna) and Peter Steinlein (Institute of Molecular Pathology, Vienna) as well as Sebastian Maurer-Stroh (Institute of Molecular Pathology, Vienna) for helpful literature suggestions. G.N. and F.E. are grateful for generous support from Boehringer Ingelheim. This research has been partially funded by the Austrian National Bank (P15037 to F.E.) and by the Fonds zur Förderung der Wissenschaftlichen Forschung Österreichs (P15037 to F.E., P15510 to J.B., P14956 to A.H.), by the Austrian Gen-AU BIN (to F.E.) and by the Austrian Ministry for Economics BMWA (to F.E.).
- Bork P, Dandekar T, Diaz-Lazcoz Y, Eisenhaber F, Huynen M, Yuan Y: Predicting function: from genes to genomes and back. J Mol Biol. 1998, 283: 707-725. 10.1006/jmbi.1998.2144.PubMedView ArticleGoogle Scholar
- Eisenhaber B, Bork P, Eisenhaber F: Prediction of potential GPI-modification sites in proprotein sequences. J Mol Biol. 1999, 292: 741-758. 10.1006/jmbi.1999.3069.PubMedView ArticleGoogle Scholar
- Eisenhaber B, Bork P, Eisenhaber F: Post-translational GPI lipid anchor modification of proteins in kingdoms of life: analysis of protein sequence data from complete genomes. Protein Eng. 2001, 14: 17-25. 10.1093/protein/14.1.17.PubMedView ArticleGoogle Scholar
- Maurer-Stroh S, Eisenhaber B, Eisenhaber F: Amino-terminal N-myristoylation of proteins: prediction of substrate proteins from amino acid sequence. J Mol Biol. 2002, 317: 541-557. 10.1006/jmbi.2002.5426.PubMedView ArticleGoogle Scholar
- Maurer-Stroh S, Eisenhaber F: Myristoylation of viral and bacterial proteins. Trends Microbiol. 2004, 12: 178-185. 10.1016/j.tim.2004.02.006.PubMedView ArticleGoogle Scholar
- Maurer-Stroh S, Gouda M, Novatchkova M, Schleiffer A, Schneider G, Sirota FL, Wildpaner M, Hayashi N, Eisenhaber F: MYRbase: analysis of genome-wide glycine myristoylation enlarges the functional spectrum of eukaryotic myristoylated proteins. Genome Biol. 2004, 5: R21-10.1186/gb-2004-5-3-r21.PubMedPubMed CentralView ArticleGoogle Scholar
- Neuberger G, Maurer-Stroh S, Eisenhaber B, Hartig A, Eisenhaber F: Prediction of peroxisomal targeting signal 1 containing proteins from amino acid sequence. J Mol Biol. 2003, 328: 581-592. 10.1016/S0022-2836(03)00319-X.PubMedView ArticleGoogle Scholar
- PTS1 prediction of Swissprot 40 entries. [http://mendel.imp.univie.ac.at/mendeljsp/sat/pts1/swissPred.jsp]
- Titorenko VI, Rachubinski RA: The peroxisome: orchestrating important developmental decisions from inside the cell. J Cell Biol. 2004, 164: 641-645. 10.1083/jcb.200312081.PubMedPubMed CentralView ArticleGoogle Scholar
- Gould SG, Keller GA, Subramani S: Identification of a peroxisomal targeting signal at the carboxy terminus of firefly luciferase. J Cell Biol. 1987, 105: 2923-2931. 10.1083/jcb.105.6.2923.PubMedView ArticleGoogle Scholar
- Gould SJ, Keller GA, Hosken N, Wilkinson J, Subramani S: A conserved tripeptide sorts proteins to peroxisomes. J Cell Biol. 1989, 108: 1657-1664. 10.1083/jcb.108.5.1657.PubMedView ArticleGoogle Scholar
- Lametschwandtner G, Brocard C, Fransen M, Van Veldhoven P, Berger J, Hartig A: The difference in recognition of terminal tripeptides as peroxisomal targeting signal 1 between yeast and human is due to different affinities of their receptor Pex5p to the cognate signal and to residues adjacent to it. J Biol Chem. 1998, 273: 33635-33643. 10.1074/jbc.273.50.33635.PubMedView ArticleGoogle Scholar
- Neuberger G, Maurer-Stroh S, Eisenhaber B, Hartig A, Eisenhaber F: Motif refinement of the peroxisomal targeting signal 1 and evaluation of taxon-specific differences. J Mol Biol. 2003, 328: 567-579. 10.1016/S0022-2836(03)00318-8.PubMedView ArticleGoogle Scholar
- Eisenhaber F, Eisenhaber B, Kubina W, Maurer-Stroh S, Neuberger G, Schneider G, Wildpaner M: Prediction of lipid posttranslational modifications and localization signals from protein sequences: big-Pi, NMT and PTS1. Nucleic Acids Res. 2003, 31: 3631-3634. 10.1093/nar/gkg537.PubMedPubMed CentralView ArticleGoogle Scholar
- Boeckmann B, Bairoch A, Apweiler R, Blatter MC, Estreicher A, Gasteiger E, Martin MJ, Michoud K, O'Donovan C, Phan I, et al: The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003. Nucleic Acids Res. 2003, 31: 365-370. 10.1093/nar/gkg095.PubMedPubMed CentralView ArticleGoogle Scholar
- Distel B, Gould SJ, Voorn-Brouwer T, van der BM, Tabak HF, Subramani S: The carboxyl-terminal tripeptide serine-lysine-leucine of firefly luciferase is necessary but not sufficient for peroxisomal import in yeast. New Biol. 1992, 4: 157-165.PubMedGoogle Scholar
- Kragler F, Langeder A, Raupachova J, Binder M, Hartig A: Two independent peroxisomal targeting signals in catalase A of Saccharomyces cerevisiae. J Cell Biol. 1993, 120: 665-673. 10.1083/jcb.120.3.665.PubMedView ArticleGoogle Scholar
- Oefner C, D'Arcy A, Winkler FK: Crystal structure of human dihydrofolate reductase complexed with folate. Eur J Biochem. 1988, 174: 377-385.PubMedView ArticleGoogle Scholar
- Oatey PB, Lumb MJ, Jennings PR, Danpure CJ: Context dependency of the PTS1 motif in human alanine: glyoxylate aminotransferase 1. Ann NY Acad Sci. 1996, 804: 652-653.PubMedView ArticleGoogle Scholar
- Oatey PB, Lumb MJ, Danpure CJ: Molecular basis of the variable mitochondrial and peroxisomal local ization of alanine-glyoxylate aminotransferase. Eur J Biochem. 1996, 241: 374-385. 10.1111/j.1432-1033.1996.00374.x.PubMedView ArticleGoogle Scholar
- Vocadlo DJ, Davies GJ, Laine R, Withers SG: Catalysis by hen egg-white lysozyme proceeds via a covalent intermediate. Nature. 2001, 412: 835-838. 10.1038/35090602.PubMedView ArticleGoogle Scholar
- Kunze M, Kragler F, Binder M, Hartig A, Gurvitz A: Targeting of malate synthase 1 to the peroxisomes of Saccharomyces cerevisiae cells depends on growth on oleic acid medium. Eur J Biochem. 2002, 269: 915-922. 10.1046/j.0014-2956.2001.02727.x.PubMedView ArticleGoogle Scholar
- Holbrook JD, Birdsey GM, Yang Z, Bruford MW, Danpure CJ: Molecular adaptation of alanine:glyoxylate aminotransferase targeting in primates. Mol Biol Evol. 2000, 17: 387-400.PubMedView ArticleGoogle Scholar
- Ueda T, Yamaguchi M, Uchimiya H, Nakano A: Ara6, a plant-unique novel type Rab GTPase, functions in the endocytic pathway of Arabidopsis thaliana. EMBO J. 2001, 20: 4730-4741. 10.1093/emboj/20.17.4730.PubMedPubMed CentralView ArticleGoogle Scholar
- Johnson MS, Johansson JM, Svensson PA, Aberg MA, Eriksson PS, Carlsson LM, Carlsson B: Interaction of scavenger receptor class B type I with peroxisomal targeting receptor Pex5p. Biochem Biophys Res Commun. 2003, 312: 1325-1334. 10.1016/j.bbrc.2003.11.076.PubMedView ArticleGoogle Scholar
- Mohan KV, Som I, Atreya CD: Identification of a type 1 peroxisomal targeting signal in a viral protein and demonstration of its targeting to the organelle. J Virol. 2002, 76: 2543-2547. 10.1128/jvi.76.5.2543-2547.2002.PubMedPubMed CentralView ArticleGoogle Scholar
- Eisenhaber B, Eisenhaber F, Maurer-Stroh S, Neuberger G: Prediction of sequence signals for lipid post-translational modifications: insights from case studies. Proteomics. 2004, 4: 1614-1625. 10.1002/pmic.200300781.PubMedView ArticleGoogle Scholar
- Eisenhaber F, Eisenhaber B, Maurer-Stroh S: Prediction of post-translational modifications from amino acid sequence: problems, pitfalls, methodological hints. Bioinformatics and Genomes: Current Perspectives. Edited by: Andrade MM. 2003, Wymondham: Horizon Scientific Press, 81-105.Google Scholar
- MacLeod KJ, Fuller RS, Scholten JD, Ahn K: Conserved cysteine and tryptophan residues of the endothelin-converting enzyme-1 CXAW motif are critical for protein maturation and enzyme activity. J Biol Chem. 2001, 276: 30608-30614. 10.1074/jbc.M103928200.PubMedView ArticleGoogle Scholar
- Bannasch D, Mehrle A, Glatting KH, Pepperkok R, Poustka A, Wiemann S: LIFEdb: a database for functional genomics experiments integrating information from external sources, and serving as a sample tracking system. Nucleic Acids Res. 2004, 32 (Database issue): D505-D508. 10.1093/nar/gkh022.View ArticleGoogle Scholar
- Kumar A, Agarwal S, Heyman JA, Matson S, Heidtman M, Piccirillo S, Umansky L, Drawid A, Jansen R, Liu Y, et al: Subcellular localization of the yeast proteome. Genes Dev. 2002, 16: 707-719. 10.1101/gad.970902.PubMedPubMed CentralView ArticleGoogle Scholar
- Brocard C, Lametschwandtner G, Koudelka R, Hartig A: Pex14p is a member of the protein linkage map of Pex5p. EMBO J. 1997, 16: 5491-5500. 10.1093/emboj/16.18.5491.PubMedPubMed CentralView ArticleGoogle Scholar
- Keller GA, Krisans S, Gould SJ, Sommer JM, Wang CC, Schliebs W, Kunau W, Brody S, Subramani S: Evolutionary conservation of a microbody targeting signal that targets proteins to peroxisomes, glyoxysomes, and glycosomes. J Cell Biol. 1991, 114: 893-904. 10.1083/jcb.114.5.893.PubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.