Adaptive encoding neural networks for the recognition of human signal peptide cleavage sites.

MOTIVATION: Data representation and encoding are essential for classification of protein sequences with artificial neural networks (ANN). Biophysical properties are appropriate for low dimensional encoding of protein sequence data. However, in general there is no a priori knowledge of the relevant properties for extraction of representative features. RESULTS: An adaptive encoding artificial neural network (ACN) for recognition of sequence patterns is described. In this approach parameters for sequence encoding are optimized within the same process as the weight vectors by an evolutionary algorithm. The method is applied to the prediction of signal peptide cleavage sites in human secretory proteins and compared with an established predictor for signal peptides. CONCLUSION: Knowledge of physico-chemical properties is not necessary for training an ACN. The advantage is a low dimensional data representation leading to computational efficiency, easy evaluation of the detected features, and high prediction accuracy. AVAILABILITY: A cleavage site prediction server is located at the Humboldt University http://itb.biologie.hu-berlin.de/ approximately jo/sig-cleave/ACNpredictor.cgi CONTACT: jo@itb.hu-berlin.de; berndj@zedat.fu-berlin.de

Visualizing and quantifying protein secretion using a Renilla luciferase-GFP fusion protein.

We have shown previously that an engineered form of Renilla luciferase (SRUC) can be secreted as a functional enzyme by mammalian cells, and that fusing wild-type Renilla luciferase with the green fluorescent protein from Aequorea victoria (GFP) yields a chimeric protein retaining light-emission properties similar to that of unfused Renilla luciferase and GFP. In the work presented here, SRUC was fused with GFP to determine whether it could be used to both visualize and quantify protein secretion in mammalian cells. Simian COS-7 and Chinese hamster ovary (CHO) cells were transiently transfected with gene constructs encoding a secreted or an intracellular version of a Renilla luciferase-GFP fusion protein. Renilla luciferase activity was measured from COS-7 cell lysates and culture media, and GFP activity was detected in CHO cells using fluorescence microscopy. Data indicated that the SRUC-GFP fusion protein was secreted as a chimeric protein that had both Renilla luciferase and GFP activity. This fusion protein could be a useful marker for the study of protein secretion in mammalian cells.

Demonstration by mass spectrometry that purified native Treponema pallidum rare outer membrane protein 1 (Tromp1) has a cleaved signal peptide.

Purified native Tromp1 was subjected to mass spectrometric analysis in order to determine conclusively whether this protein possesses a cleaved or uncleaved signal peptide. The molecular masses of Tromp1, three Treponema pallidum lipoproteins, and a bovine serum albumin (BSA) control were determined by matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry. The molecular masses of all of the T. pallidum lipoproteins and BSA were within 0.7% of their respective calculated masses. The molecular mass of Tromp1 was 31,510 Da, which is consistent with a signal-less form of Tromp1, given a calculated mass of unprocessed Tromp1 of 33, 571 Da, a difference of 2,061 Da (a 6.5% difference). Purified native Tromp1 was also subjected to MALDI-TOF analysis in comparison to recombinant Tromp1 following cyanogen bromide cleavage, which further confirmed the identity of Tromp1 and showed that native Tromp1 was not degraded at the carboxy terminus. These studies confirm that Tromp1 is processed and does not contain an uncleaved signal peptide as previously reported.

Different sequence patterns in signal peptides from mycoplasmas, other gram-positive bacteria, and Escherichia coli: a multivariate data analysis.

Signal peptides are essential N-terminal extensions in export proteins, and have a positively charged N-terminus, a hydrophobic central core, and a C-terminal cleavage region. They interact in a consecutive manner with different accessory proteins during the secretion process. Potential patterns or periodicity in the amino acid (aa) sequence were searched, using multivariate techniques, for a large number of signal peptides from mollicutes (mycoplasmas), other Gram-positive bacteria, and Escherichia coli. Mollicutes signal peptides were significantly different from the E. coli and Gram-positive ones by their N-terminal charge, peptide length, and especially, unique periodicities of side chain hydrophobicity and volumes. Their lipoprotein signal peptides were longer than for any other bacteria. Significant differences were also recorded between the other bacterial peptide groups. Specific aa patterns were more related within the signal peptides from several groups of secreted bacillus enzymes, than for all signal peptides from one bacillus species. In E. coli, signal peptides from proteins routed for the various destinations revealed significant and compartment-specific sequence patterns not evident by other methods. This was substantiated from a large number of signal peptide secretion mutants for the E. coli periplasmic space. It is proposed that the differences in aa patterns and side-chain properties are related to the secondary structure sidedness and topology of the signal peptides, and important for specific interactions during the secretion process.

Machine learning approaches for the prediction of signal peptides and other protein sorting signals.

Prediction of protein sorting signals from the sequence of amino acids has great importance in the field of proteomics today. Recently, the growth of protein databases, combined with machine learning approaches, such as neural networks and hidden Markov models, have made it possible to achieve a level of reliability where practical use in, for example automatic database annotation is feasible. In this review, we concentrate on the present status and future perspectives of SignalP, our neural network-based method for prediction of the most well-known sorting signal: the secretory signal peptide. We discuss the problems associated with the use of SignalP on genomic sequences, showing that signal peptide prediction will improve further if integrated with predictions of start codons and transmembrane helices. As a step towards this goal, a hidden Markov model version of SignalP has been developed, making it possible to discriminate between cleaved signal peptides and uncleaved signal anchors. Furthermore, we show how SignalP can be used to characterize putative signal peptides from an archaeon, Methanococcus jannaschii. Finally, we briefly review a few methods for predicting other protein sorting signals and discuss the future of protein sorting prediction in general.

Signal sequence recognition in posttranslational protein transport across the yeast ER membrane.

We have analyzed how the signal sequence of prepro-alpha-factor is recognized during the first step of posttranslational protein transport into the yeast endoplasmic reticulum. Cross-linking studies indicate that the signal sequence interacts in a Kar2p- and ATP-independent reaction with Sec61p, the multispanning membrane component of the protein-conducting channel, by intercalation into transmembrane domains 2 and 7. While bound to Sec61p, the signal sequence forms a helix that is contacted on one side by Sec62p and Sec71p. The binding site is located at the interface of the protein channel and the lipid bilayer. Signal sequence recognition in cotranslational translocation in mammals appears to occur similarly. These results suggest a general mechanism by which the signal sequence could open the channel for polypeptide transport.

ERAB contains a putative noncleavable signal peptide.

The 42-residue amyloid beta protein (Abeta42) has been shown to be toxic to neurons and is believed to play a key causative role in Alzheimer's disease (AD). A search for Abeta binding proteins that can mediate its toxicity resulted in the identification of the endoplasmic-reticulum (ER) associated Abeta binding protein (ERAB) which was also shown to be involved in Abeta induced apoptosis. The primary report indicated that a signal sequence is absent in ERAB suggesting that it is bound to the cytoplasmic aspect of cellular membranes. Abeta is generated in the lumen of secretory organelles and released into the medium resulting in its separation from ERAB by a membrane barrier. After computer analysis of the ERAB sequence, we have detected a putative signal peptide that can direct the protein into the secretory pathway. This signal sequence was found in human, rodent, and bovine ERAB suggesting that it is a type II integral membrane protein in vertebrates. This topology can explain the binding of Abeta to ERAB. Our finding that an integral membrane form of ERAB can bind to Abeta in the lumen of transport vesicles and other cytoplasmic receptors provides a basis for understanding its role in AD.

Nuclear localization signal peptides enhance cationic liposome-mediated gene therapy.

The use of genes as therapeutic drugs will likely involve non-viral delivery systems. While traditionally less effective for gene expression, the advantages of a non-viral delivery system include ease of production, lower toxicity, and no risk of infection. However, most non-viral systems do not incorporate a mechanism for gene transport into the nucleus. Nuclear localization signal peptides can combine the increased expression of viral delivery systems with the safety and ease of preparation of non-viral delivery systems. A novel non-viral delivery vehicle consisting of a conglomerate of a synthetic nuclear localization signal peptide derived from the SV40 virus, a luciferase encoding PGL3 plasmid, and a cationic lipid DOTAP:DOPE (1:1 w/w) liposome was transfected into SKnSH mammalian neuroblastoma cells. A three-fold increase in luciferase expression was seen with the delivery system containing a NLS peptide over cationic liposome controls. Examination of the factors that limit the rate of transgene expression can potentially lead to the discovery of new ways to improve the efficiency and efficacy of nonviral methods of gene therapy.

A novel plasmid recombination mechanism of the marine cyanobacterium Synechococcus sp. PCC7002.

We describe a novel mechanism of site-specific recombination in the unicellular marine cyanobacterium Synechococcus sp. PCC7002. The specific recombination sites on the smallest plasmid pAQ1 were localized by studying the properties of pAQ1-derived shuttle-vectors. We found that a palindromic element, the core sequence of which is G(G/A)CGATCGCC, functions as a resolution site for site-specific plasmid recombination. Furthermore, site-directed mutagenesis analysis of the element show that the site-specific recombination in the cyanobacterium requires sequence specificity, symmetry in the core sequence and, in part, the spacing between the elements. Interestingly, this element is over-represented not only in pAQ1 and in the genome of the cyanobacterium, but also in the accumulated cyanobacterial sequences from Synechococcus sp. PCC6301, PCC7942, vulcanus and Synechocystis sp. PCC6803 within GenBank and EMBL databases. Thus, these findings strongly suggest that the site-specific recombination mechanism based on the palindromic element should be common in these cyanobacteria.

Human cDNA encoding a novel TGF-beta superfamily protein highly expressed in placenta.

Recently, we developed a simple method for detecting a secretory signal sequence encoded by a cDNA fragment. In this study, we used this method to select cDNA clones encoding secretory proteins from a human full-length cDNA library. Full-sequencing analysis of the candidate clones revealed that one clone encoded a novel TGF-beta superfamily protein. The clone encodes a protein of 308 amino acids of which the C-terminal region shows a characteristic feature of TGF-beta superfamily proteins: seven conserved cysteine residues at the C-terminal preceded by a putative processing site composed of a basic amino acid repeat. The corresponding transcripts are highly expressed in the placenta, so the novel protein may play an important role in reproduction.

Localization and characterization of inclusion bodies in recombinant Escherichia coli cells overproducing penicillin G acylase.

Various concentrations of isopropyl beta-D-thiogalactopyranoside (IPTG) were used to induce production of the enzyme penicillin G acylase by recombinant Escherichia coli harboring plasmid pQEA11. The plasmid pQEA11 carries a wild-type pga gene, which is under the control of the tac promoter and lacIq. At low IPTG concentrations (0.025-0.1 mM), enzyme activity increased with increasing IPTG concentrations. At higher IPTG concentrations (0.2 and 0.5 mM), enzyme activity declined progressively. Examination of induced recombinant E. coli cells by transmission electron microscopy showed the presence of only periplasmic inclusion bodies at low IPTG concentrations (up to 0.1 mM) and both periplasmic and cytoplasmic inclusion bodies at high IPTG concentrations (0.2 mM and 0.5 mM). Results from sodium dodecyl sulfate/polyacrylamide gel electrophoresis and immunoblots of whole-cell proteins, membrane proteins and inclusion body proteins in these cells indicated that cytoplasmic inclusion bodies constituted an accumulation of preproenzyme (i.e., precursor polypeptide containing a signal peptide) and that periplasmic inclusion bodies constituted an accumulation of proenzyme (i.e., precursor polypeptide lacking a signal peptide).

Characteristics of nucleotide sequences flanking the trans-spliced leader SL1 exon in Dirofilaria immitis, Brugia malayi, and Brugia pahangi.

Nucleotide sequences surrounding the trans-spliced leader SL1 exon in the 5S rRNA gene spacer regions of Dirofilaria immitis, Brugia malayi, and B. pahangi were determined after PCR amplification, aligned with the genus Onchocerca for comparison, and used for the prediction of secondary structures. The nucleotide sequence of this region in B. pahangi was first shown in the present study. Hypothetical secondary structures of the spacer region suggested that the SL1 transcript is capable to form a stable stem-loop structure which may render transposition of the SL1 sequence to mRNA molecules. A homologous sequence to Sm-binding site was assigned on a bulge loop. No significant difference was observed in adult worms of D. immitis irrespective of sex or location. No difference was apparent between the two species in genus Brugia.

The sequence NPFXD defines a new class of endocytosis signal in Saccharomyces cerevisiae.

The yeast membrane protein Kex2p uses a tyrosine-containing motif within the cytoplasmic domain for localization to a late Golgi compartment. Because Golgi membrane proteins mislocalized to the plasma membrane in yeast can undergo endocytosis, we examined whether the Golgi localization sequence or other sequences in the Kex2p cytoplasmic domain mediate endocytosis. To assess endocytic function, the Kex2p cytoplasmic domain was fused to an endocytosis-defective form of the alpha-factor receptor. Ste2p. Like intact Ste2p, the chimeric protein, Stex22p, undergoes rapid endocytosis that is dependent on clathrin and End3p. Uptake of Stex22p does not require the Kex2p Golgi localization motif. Instead, the sequence NPFSD, located 37 amino acids from the COOH terminus, is essential for Stex22p endocytosis. Internalization was abolished when the N, P, or F residues were converted to alanine and severely impaired upon conversion of D to A. NPFSD restored uptake when added to the COOH terminus of an endocytosis-defective Ste2p chimera lacking lysine-based endocytosis signals present in wild-type Ste2p. An NPF sequence is present in the cytoplasmic domain of the a-factor receptor, Ste3p. Mutation of this sequence prevented pheromone-stimulated endocytosis of a truncated form of Ste3p. Our results identify NPFSD as a clathrin-dependent endocytosis signal that is distinct from the aromatic amino acid-containing Golgi localization motif and lysine-based, ubiquitin-dependent endocytosis signals in yeast.

Analysis of the C-terminal secretion signal of the Rhizobium leguminosarum nodulation protein NodO; a potential system for the secretion of heterologous proteins during nodule invasion.

We used deletions to analyze the domains required for secretion of the Rhizobium leguminosarum bv. viciae nodulation protein, NodO, by the sec-independent pathway. Deletion of the C-terminal 24 amino-acids (residues 261 to 284) reduced secretion by at least 95%. A monoclonal antibody that recognizes the C-terminal domain of NodO was used to identify four nested deletions that retained the C-terminal 24 residues of NodO but had lost up to 133 residues (amino acids 128 to 259); all four proteins were secreted into the growth medium with an efficiency between 50 and 90% of normal. A deleted derivative of NodO that retained residues 1 to 21 and 167 to 284 (and therefore lacked most of the N-terminal Ca(2+)-binding domain) was secreted at around 80% of normal efficiency. Taken together, these observations indicate that the C-terminal 24 amino acids are sufficient for NodO secretion although the region adjacent to this domain appears to affect secretion efficiency. A derivative of the Escherichia coli alkaline phosphatase (phoA) gene was cloned into two derivatives of nodO such that PhoA (lacking the N-terminal transit peptide) was in-frame at both ends, with the C terminus fused to either the last 24 or 50 amino acids of NodO. These fusion proteins were secreted at 40 and 80% of the wild-type level, respectively, and the larger of the two retained alkaline phosphatase activity. A hybrid protein, containing E. coli beta-glucuronidase (GUS) fused to the N terminus of NodO, was not secreted, and it reduced the levels of wild-type NodO secreted by R. leguminosarum bv. viciae. The nature of the NodO C-terminal secretion signal is discussed with regard to its use as a delivery system for heterologous proteins useful for investigating the Rhizobium-legume interaction.

Translational machinery in dendrites of hippocampal neurons in culture.

In neurons, several mRNAs are selectively delivered to dendritic domains where they are presumably translated by local protein synthetic machinery. Although electron microscopy has identified polyribosomes in dendrites, in particular in postsynaptic dendritic compartments, the functional composition of the local protein synthetic apparatus and the scope of its translational capacity have not been analyzed. To ascertain the translational competence of dendrites, we have probed hippocampal neurons in primary culture for various integral and associated factors of the translational apparatus. We report here that dendrites of such neurons are equipped with a spectrum of translational machinery components, including ribosomes, tRNAs, initiation and elongation factors, and elements of the cotranslational signal recognition mechanism. These components are differentially and nonuniformly distributed in dendritic arbors. Their dendritic location illustrates the soma-independent potential of dendrites to synthesize selected proteins in local domains.

Novel lantibiotics and their pre-peptides.

In recent years there has been a considerable increase in studies of bactericidal peptides produced by Gram-positive bacteria, with particular emphasis upon their potential application as food preservatives. A number of these peptides contain lanthionine and other post-translationally modified amino acid residues. The lanthionine-containing molecules (lantibiotic) appear to have evolved in two quite different lineages, type A and type B. This mini-review introduces the reader to several of the more recently described type A lantibiotics for which relatively detailed biochemical and/or genetic data has already been gathered. A wider diversity of compounds of type A lantibiotics has been described in the recent years. Novel features of some of the more recently described type A lantibiotics to be reported in this review include: a) New modifications such as D-Ala and 2-hydroxypropionyl residues, both derived from serine. b) Different types of pre-lantibiotic leader sequences. c) The apparent requirement for different numbers and types of genes for synthesis of some active type A lantibiotics. d) Cytolysin functions as both a hemolysin and a bacteriocin. e) One of the newly-described lantibiotics (lactocin S) does not have any net charge at neutral pH another (carnocin UI49) is the largest of the lantibiotics discovered and the killing action of another (cytolysin) has been shown to be depend on the interaction of two peptides.

Sequence requirement for nucleolar localization of rat ribosomal protein L31.

Nucleolar localization of the fusion protein of rat ribosomal protein L31 and beta-galactosidase was investigated by immunocytochemical means, using the ribosomal protein deletion or substitution mutants that were transiently expressed in COS-1 cells. The signal responsible for nucleolar localization is encoded by the amino acid residues 87 to 92, RLSRKR, located near the C terminus of the ribosomal protein. Mutation of residues 87R and 90R prevented nucleolar localization, whereas mutations including 90R prevented nuclear and nucleolar localization. Further mutations in the sequence revealed that the tetrapeptide RLSR, which was amenable to substitutions at the L and S positions, is critical for nucleolar localization.

Sequence and product analyses of the four genes downstream from the fimbrilin gene (fimA) of the oral anaerobe Porphyromonas gingivalis.

The downstream DNA region of the fimbrilin gene (fimA), which encodes the major subunit protein of Porphyromonas gingivalis fimbriae, was fully sequenced. Gene products, expressed from this region in Escherichia coli, were purified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and their partial amino acid sequences were determined to verify open reading frames (ORFs) found in the region by DNA sequencing. Four ORFs, designated ORF1, ORF2, ORF3 and ORF4, were found in the 5.8-kb PstI fragment downstream from fimA, which was previously cloned and partially characterized by Yoshimura, Takahashi, Hibi, Takasawa, Kato, and Dickinson (Infect. Immun. 61: 5181-5189, 1993). The direction of transcription of all the ORFs was the same as that of fimA. The 50 and 80 kDa encoded proteins, ORF2 and ORF3, respectively, have been reported to be minor components associated with fimbriae. The 15 and 19 kDa proteins, ORF1 and ORF4, respectively, have been expressed in E. coli but not identified in P. gingivalis. However, all the gene products of the ORFs, expressed in E. coli, appeared to contain intact signal peptides based on their N-terminal amino acid sequences.

The cytoplasmic and N-terminal transmembrane domains of cytochrome P450 contain independent signals for retention in the endoplasmic reticulum.

Microsomal cytochrome P450 is inserted into the membrane of the endoplasmic reticulum (ER) by its N-terminal signal/anchor sequence which also functions as an ER retention signal. To analyze further potential retention signals of cytochrome P450, topological domains of cytochrome P450 2C1 or 2C2, epidermal growth factor receptor, a plasma membrane protein, and bacterial alkaline phosphatase, a secreted protein were exchanged. The N-terminal signal/anchor of cytochrome P450 2C1 functioned as an ER retention signal when placed at the N terminus of several reporter proteins but not when fused at the C terminus of the extracellular domain of epidermal growth factor receptor, with or without a heterologous cytoplasmic domain. Chimeric proteins in which the cytoplasmic domain of cytochrome P450 2C2 was substituted for that of epidermal growth factor receptor were retained in the ER indicating that an independent retention signal is present in the cytoplasmic part of cytochrome P450 2C2. These chimeras were enzymatically active which argues against misfolding as the primary cause of retention. The ER retention signal of the cytoplasmic domain could not be localized to a single amino acid segment by deletion analysis. These results show that cytochrome P450 2C2 contains redundant, complex ER retention signals in its cytoplasmic and N-terminal hydrophobic domains and that the function of the N-terminal signal is context-dependent.