Consensus predictions of membrane protein topology

We have explored the possibility that consensus predictions of membrane protein topology might provide a means to estimate the reliability of a predicted topology. Using five current topology prediction methods and a test set of 60 Escherichia coli inner membrane proteins with experimentally determined topologies, we find that prediction performance varies strongly with the number of methods that agree, and that the topology of nearly half of all E. coli inner membrane proteins can be predicted with high reliability (>90% correct predictions) by a simple majority-vote approach.     

Comparative analysis of amino acid distributions in integral membrane proteins from 107 genomes

We have performed a comparative analysis of amino acid distributions in predicted integral membrane proteins from a total of 107 genomes. A procedure for identification of membrane spanning helices was optimized on a homology-reduced data set of 170 multi-spanning membrane proteins with experimentally determined topologies. The optimized method was then used for extraction of highly reliable partial topologies from all predicted membrane proteins in each genome, and the average biases in amino acid distributions between loops on opposite sides of the membrane were calculated. The results strongly support the notion that a biased distribution of Lys and Arg residues between cytoplasmic and extra-cytoplasmic segments (the positive-inside rule) is present in most if not all organisms.     

Alpha helical trans-membrane proteins: Enhanced prediction using a Bayesian approach

Membrane proteins, which constitute approximately 20% of most genomes, are poorly tractable targets for experimental structure determination, thus analysis by prediction and modelling makes an important contribution to their on-going study. Membrane proteins form two main classes: alpha helical and beta barrel trans-membrane proteins. By using a method based on Bayesian Networks, which provides a flexible and powerful framework for statistical inference, we addressed alpha-helical topology prediction. This method has accuracies of 77.4% for prokaryotic proteins and 61.4% for eukaryotic proteins. The method described here represents an important advance in the computational determination of membrane protein topology and offers a useful, and complementary, tool for the analysis of membrane proteins for a range of applications.     

Reliability measures for membrane protein topology prediction algorithms

We have developed reliability scores for five widely used membrane protein topology prediction methods, and have applied them both on a test set of 92 bacterial plasma membrane proteins with experimentally determined topologies and on all predicted helix bundle membrane proteins in three fully sequenced genomes: Escherichia coli, Saccharomyces cerevisiae and Caenorhabditis elegans. We show that the reliability scores work well for the TMHMM and MEMSAT methods, and that they allow the probability that the predicted topology is correct to be estimated for any protein. We further show that the available test set is biased towards high-scoring proteins when compared to the genome-wide data sets, and provide estimates for the expected prediction accuracy of TMHMM across the three genomes. Finally, we show that the performance of TMHMM is considerably better when limited experimental information (such as the in/out location of a protein's C terminus) is available, and estimate that at least ten percentage points in overall accuracy in whole-genome predictions can be gained in this way.     

Prediction of Membrane Protein Topology Utilizing Multiple Sequence Alignments

A technique for prediction of protein membrane toplogy (intra- and extraceullular sidedness) has been developed. Membrane-spanning segments are first predicted using an algorithm based upon multiply aligned amino acid sequences. The compositional differences in the protein segments exposed at each side of the membrane are then investigated. The ratios are calculated for Asn, Asp, Gly, Phe, Pro, Trp, Tyr, and Val, mostly found on the extracellular side, and for Ala, Arg, Cys, and Lys, mostly occurring on the intracellular side. The consensus over these 12 residue distributions is used for sidedness prediction. The method was developed with a set of 42 protein families for which all but one were correctly predicted with the new algorithm. This represents an improvement over previous techniques. The new method, applied to a set of 12 membrane protein families different from the test set and with recently determined topologies, performed well, with 11 of 12 sidedness assignments agreeing with experimental results. The method has also been applied to several membrane protein families for which the topology has yet to be determined. An electronic prediction service is available at the E-mail address tmap@embl-heidelberg.de and on WWW via http://www.emblheidelberg.de.     

High-throughput production of prokaryotic membrane proteins

Membrane proteins constitute ~30% of prokaryotic and eukaryotic genomes but comprise a small fraction of the entries in protein structural databases. A number of features of membrane proteins render them challenging targets for the structural biologist, among which the most important is the difficulty in obtaining sufficient quantities of purified protein. We are exploring procedures to express and purify large numbers of prokaryotic membrane proteins. A set of 280 membrane proteins from Escherichia coli and Thermotoga maritima, a thermophile, was cloned and tested for expression in Escherichia coli. Under a set of standard conditions, expression could be detected in the membrane fraction for approximately 30% of the cloned targets. About 22 of the highest expressing membrane proteins were purified, typically in just two chromatographic steps. There was a clear correlation between the number of predicted transmembrane domains in a given target and its propensity to express and purify. Accordingly, the vast majority of successfully expressed and purified proteins had six or fewer transmembrane domains. We did not observe any clear advantage to the use of thermophilic targets. Two of the purified membrane proteins formed crystals. By comparison with protein production efforts for soluble proteins, where ∼70% of cloned targets express and ∼25% can be readily purified for structural studies [Christendat et al. (2000) Nat. Struct. Biol., 7, 903], our results demonstrate that a similar approach will succeed for membrane proteins, albeit with an expected higher attrition rate.     

A guideline to proteome-wide α-helical membrane protein topology predictions

For current state-of-the-art methods, the prediction of correct topology of membrane proteins has been reported to be above 80%. However, this performance has only been observed in small and possibly biased data sets obtained from protein structures or biochemical assays. Here, we test a number of topology predictors on an "unseen" set of proteins of known structure and also on four "genome-scale" data sets, including one recent large set of experimentally validated human membrane proteins with glycosylated sites. The set of glycosylated proteins is also used to examine the ability of prediction methods to separate membrane from nonmembrane proteins. The results show that methods utilizing multiple sequence alignments are overall superior to methods that do not. The best performance is obtained by TOPCONS, a consensus method that combines several of the other prediction methods. The best methods to distinguish membrane from nonmembrane proteins belong to the "Phobius" group of predictors. We further observe that the reported high accuracies in the smaller benchmark sets are not quite maintained in larger scale benchmarks. Instead, we estimate the performance of the best prediction methods for eukaryotic membrane proteins to be between 60% and 70%. The low agreement between predictions from different methods questions earlier estimates about the global properties of the membrane proteome. Finally, we suggest a pipeline to estimate these properties using a combination of the best predictors that could be applied in large-scale proteomics studies of membrane proteins.     

Prediction by a neural network of outer membrane beta-strand protein topology.

An artificial neural network (NN) was trained to predict the topology of bacterial outer membrane (OM) beta-strand proteins. Specifically, the NN predicts the z-coordinate of Calpha atoms in a coordinate frame with the outer membrane in the xy-plane, such that low z-values indicate periplasmic turns, medium z-values indicate transmembrane beta-strands, and high z-values indicate extracellular loops. To obtain a training set, seven OM proteins (porins) with structures known to high resolution were aligned with their pores along the z-axis. The relationship between Calpha z-values and topology was thereby established. To predict the topology of other OM proteins, all seven porins were used for the training set. Z-values (topologies) were predicted for two porins with hitherto unknown structure and for OM proteins not belonging to the porin family, all with insignificant sequence homology to the training set. The results of topology prediction compare favorably with experimental topology data.     

Comparative Analysis of the Ribosomal Componentsof the Hydrogenosome-Containing Protist, Trichomonas vaginalis

The ribosomes of the amitochondriate but hydrogenosome-containing protist lineage, the trichomonads, have previously been reported to be prokaryotic or primitive eukaryotic, based on evidence that they have a 70S sedimentation coefficient and a small number of proteins, similar to prokaryotic ribosomes. In order to determine whether the components of the trichomonad ribosome indeed differ from those of typical eukaryotic ribosomes, the ribosome of a representative trichomonad, Trichomonas vaginalis, was characterized. The sedimentation coefficient of the T. vaginalis ribosome was smaller than that of Saccharomyces cerevisiae and larger than that of Escherichia coli. Based on two-dimensional PAGE analysis, the number of different ribosomal proteins was estimated to be approximately 80. This number is the same as those obtained for typical eukaryotes (approximately 80) but larger than that of E. coli (approximately 55). N-Terminal amino acid sequencing of 18 protein spots and the complete sequences of 4 ribosomal proteins as deduced from their genes revealed these sequences to display typical eukaryotic features. Phylogenetic analyses of the five ribosomal proteins currently available also clearly confirmed that the T. vaginalis sequences are positioned within a eukaryotic clade. Comparison of deduced secondary structure models of the small and large subunit rRNAs of T. vaginalis with those of other eukaryotes revealed that all helices commonly found in typical eukaryotes are present and conserved in T. vaginalis, while variable regions are shortened or lost. These lines of evidence demonstrate that the T. vaginalis ribosome has no prokaryotic or primitive eukaryotic features but is clearly a typical eukaryotic type.     

The bestrophin family of anion channels: identification of prokaryotic homologues

The human disease protein, Bestrophin-1, associated with vitelliform macular dystrophy, has recently been shown to be an integral membrane anion channel-forming protein. In this study we have recovered all bestrophin homologues from the NCBI database and analyzed their sequences using bioinformatic approaches. Eukaryotic homologues were found in animals and fungi but not in plants or protozoans, and prokaryotic homologues distantly related to the eukaryotic proteins, were identified in certain Gram-negative bacterial kingdoms but not in Gram-positive bacteria or archaea. Our analyses suggest a uniform 4 TMS topology for most of these homologues with regions of conservation overlapping and preceding the odd numbered TMSs and overlapping and following the even numbered TMSs. Well-conserved motifs were identified in both the eukaryotic and the prokaryotic homologues, and these proved to overlap, suggesting common structural and functional properties. Phylogenetic analyses revealed that the eukaryotic proteins cluster according to organismal type, and that the prokaryotic proteins sometimes (but not always) do so. This suggests that eukaryotic paralogues arose exclusively by recent gene duplication events although both early and late gene duplication events occurred in prokaryotes.     

Improved membrane protein topology prediction by domain assignments

Topology predictions for integral membrane proteins can be substantially improved if parts of the protein can be constrained to a given in/out location relative to the membrane using experimental data or other information. Here, we have identified a set of 367 domains in the SMART database that, when found in soluble proteins, have compartment-specific localization of a kind relevant for membrane protein topology prediction. Using these domains as prediction constraints, we are able to provide high-quality topology models for 11% of the membrane proteins extracted from 38 eukaryotic genomes. Two-thirds of these proteins are single spanning, a group of proteins for which current topology prediction methods perform particularly poorly.     

Comparative interactomics analysis of protein family interaction networks using PSIMAP (protein structural interactome map)

MOTIVATION: Many genomes have been completely sequenced. However, detecting and analyzing their protein-protein interactions by experimental methods such as co-immunoprecipitation, tandem affinity purification and Y2H is not as fast as genome sequencing. Therefore, a computational prediction method based on the known protein structural interactions will be useful to analyze large-scale protein-protein interaction rules within and among complete genomes. RESULTS: We confirmed that all the predicted protein family interactomes (the full set of protein family interactions within a proteome) of 146 species are scale-free networks, and they share a small core network comprising 36 protein families related to indispensable cellular functions. We found two fundamental differences among prokaryotic and eukaryotic interactomes: (1) eukarya had significantly more hub families than archaea and bacteria and (2) certain special hub families determined the topology of the eukaryotic interactomes. Our comparative analysis suggests that a very small number of expansive protein families led to the evolution of interactomes and seemed to have played a key role in species diversification. SUPPLEMENTARY INFORMATION: http://interactomics.org.     

An improved hidden Markov model for transmembrane protein detection and topology prediction and its applications to complete genomes

MOTIVATION: Knowledge of the transmembrane helical topology can help identify binding sites and infer functions for membrane proteins. However, because membrane proteins are hard to solubilize and purify, only a very small amount of membrane proteins have structure and topology experimentally determined. This has motivated various computational methods for predicting the topology of membrane proteins. RESULTS: We present an improved hidden Markov model, TMMOD, for the identification and topology prediction of transmembrane proteins. Our model uses TMHMM as a prototype, but differs from TMHMM by the architecture of the submodels for loops on both sides of the membrane and also by the model training procedure. In cross-validation experiments using a set of 83 transmembrane proteins with known topology, TMMOD outperformed TMHMM and other existing methods, with an accuracy of 89% for both topology and locations. In another experiment using a separate set of 160 transmembrane proteins, TMMOD had 84% for topology and 89% for locations. When utilized for identifying transmembrane proteins from non-transmembrane proteins, particularly signal peptides, TMMOD has consistently fewer false positives than TMHMM does. Application of TMMOD to a collection of complete genomes shows that the number of predicted membrane proteins accounts for approximately 20-30% of all genes in those genomes, and that the topology where both the N- and C-termini are in the cytoplasm is dominant in these organisms except for Caenorhabditis elegans. AVAILABILITY: http://liao.cis.udel.edu/website/servers/TMMOD/     

Comparative analysis and "expression space" coverage of the production of prokaryotic membrane proteins for structural genomics

Membrane proteins comprise up to one-third of prokaryotic and eukaryotic genomes, but only a very small number of membrane protein structures are known. Membrane proteins are challenging targets for structural biology, primarily due to the difficulty in producing and purifying milligram quantities of these proteins. We are evaluating different methods to produce and purify large numbers of prokaryotic membrane proteins for subsequent structural and functional analysis. Here, we present the comparative expression data for 37 target proteins, all of them secondary transporters, from the mesophilic organism Salmonella typhimurium and the two hyperthermophilic organisms Aquifex aeolicus and Pyrococcus furiosus in three different Escherichia coli expression vectors. In addition, we study the use of Lactococcus lactis as a host for integral membrane protein expression. Overall, 78% of the targets were successfully produced under at least one set of conditions. Analysis of these results allows us to assess the role of different variables in increasing "expression space" coverage for our set of targets. This analysis implies that to maximize the number of nonhomologous targets that are expressed, orthologous targets should be chosen and tested in two vectors with different types of promoters, using C-terminal tags. In addition, E. coli is shown to be a robust host for the expression of prokaryotic transporters, and is superior to L. lactis. These results therefore suggest appropriate strategies for high-throughput heterologous overproduction of membrane proteins.     

Consequences of the overexpression of a eukaryotic membrane protein, the human KDEL receptor, in Escherichia coli

Escherichia coli is the most widely used host for producing membrane proteins. Thus far, to study the consequences of membrane protein overexpression in E. coli, we have focussed on prokaryotic membrane proteins as overexpression targets. Their overexpression results in the saturation of the Sec translocon, which is a protein-conducting channel in the cytoplasmic membrane that mediates both protein translocation and insertion. Saturation of the Sec translocon leads to (i) protein misfolding/aggregation in the cytoplasm, (ii) impaired respiration, and (iii) activation of the Arc response, which leads to inefficient ATP production and the formation of acetate. The overexpression yields of eukaryotic membrane proteins in E. coli are usually much lower than those of prokaryotic ones. This may be due to differences between the consequences of the overexpression of prokaryotic and eukaryotic membrane proteins in E. coli. Therefore, we have now also studied in detail how the overexpression of a eukaryotic membrane protein, the human KDEL receptor, affects E. coli. Surprisingly, the consequences of the overexpression of a prokaryotic and a eukaryotic membrane protein are very similar. Strain engineering and likely also protein engineering can be used to remedy the saturation of the Sec translocon upon overexpression of both prokaryotic and eukaryotic membrane proteins in E. coli.     

A predictor of membrane class: Discriminating alpha-helical and beta-barrel membrane proteins from non-membranous proteins

Accurate protein structure prediction remains an active objective of research in bioinformatics. Membrane proteins comprise approximately 20% of most genomes. They are, however, poorly tractable targets of experimental structure determination. Their analysis using bioinformatics thus makes an important contribution to their on-going study. Using a method based on Bayesian Networks, which provides a flexible and powerful framework for statistical inference, we have addressed the alignment-free discrimination of membrane from non-membrane proteins. The method successfully identifies prokaryotic and eukaryotic alpha-helical membrane proteins at 94.4% accuracy, beta-barrel proteins at 72.4% accuracy, and distinguishes assorted non-membranous proteins with 85.9% accuracy. The method here is an important potential advance in the computational analysis of membrane protein structure. It represents a useful tool for the characterisation of membrane proteins with a wide variety of potential applications.     

Evaluation of the membrane-spanning domain of ClC-2

The ClC family of chloride channels and transporters includes several members in which mutations have been associated with human disease. An understanding of the structure-function relationships of these proteins is essential for defining the molecular mechanisms underlying pathogenesis. To date, the X-ray crystal structures of prokaryotic ClC transporter proteins have been used to model the membrane domains of eukaryotic ClC channel-forming proteins. Clearly, the fidelity of these models must be evaluated empirically. In the present study, biochemical tools were used to define the membrane domain boundaries of the eukaryotic protein, ClC-2, a chloride channel mutated in cases of idiopathic epilepsy. The membrane domain boundaries of purified ClC-2 and accessible cysteine residues were determined after its functional reconstitution into proteoliposomes, labelling using a thiol reagent and proteolytic digestion. Subsequently, the lipid-embedded and soluble fragments generated by trypsin-mediated proteolysis were studied by MS and coverage of approx. 71% of the full-length protein was determined. Analysis of these results revealed that the membrane-delimited boundaries of the N- and C-termini of ClC-2 and the position of several extramembrane loops determined by these methods are largely similar to those predicted on the basis of the prokaryotic protein [ecClC (Escherichia coli ClC)] structures. These studies provide direct biochemical evidence supporting the relevance of the prokaryotic ClC protein structures towards understanding the structure of mammalian ClC channel-forming proteins.     

Identification of membrane proteins in the hyperthermophilic archaeon pyrococcus furiosus using proteomics and prediction programs

Cell-free extracts from the hyperthermophilic archaeon Pyrococcus furiosus were separated into membrane and cytoplasmic fractions and each was analyzed by 2D-gel electrophoresis. A total of 66 proteins were identified, 32 in the membrane fraction and 34 in the cytoplasmic fraction. Six prediction programs were used to predict the subcellular locations of these proteins. Three were based on signal-peptides (SignalP, TargetP, and SOSUISignal) and three on transmembrane-spanning alpha-helices (TSEG, SOSUI, and PRED-TMR2). A consensus of the six programs predicted that 23 of the 32 proteins (72%) from the membrane fraction should be in the membrane and that all of the proteins from the cytoplasmic fraction should be in the cytoplasm. Two membrane-associated proteins predicted to be cytoplasmic by the programs are also predicted to consist primarily of transmembrane-spanning beta-sheets using porin protein models, suggesting that they are, in fact, membrane components. An ATPase subunit homolog found in the membrane fraction, although predicted to be cytoplasmic, is most likely complexed with other ATPase subunits in the membrane fraction. An additional three proteins predicted to be cytoplasmic but found in the membrane fraction, may be cytoplasmic contaminants. These include a chaperone homolog that may have attached to denatured membrane proteins during cell fractionation. Omitting these three proteins would boost the membrane-protein predictability of the models to near 80%. A consensus prediction using all six programs for all 2242 ORFs in the P. furiosus genome estimates that 24% of the ORF products are found in the membrane. However, this is likely to be a minimum value due to the programs' inability to recognize certain membrane-related proteins, such as subunits associated with membrane complexes and porin-type proteins.     

Protein topology prediction through constraint-based search and the evaluation of topological folding rules.

An algorithm for predicting protein alpha/beta-sheet topologies from secondary structure and topological folding rules (constraints) has been developed and implemented in Prolog. This algorithm (CBS1) is based on constraint satisfaction and employs forward pruned breadth-first search and rotational invariance. CBS1 showed a 37-fold increase in efficiency over an exhaustive generate and test algorithm giving the same solution for a typical sheet of five strands whose topology was predicted from secondary structure with four topological folding constraints. Prolog specifications of a range of putative protein folding rules were then used to (i) replicate published protein topology predictions and (ii) validate these rules against known protein structures of nucleotide-binding domains. This demonstrated that (i) manual techniques for topology prediction can lead to non-exhaustive search and (ii) most of these protein folding principles were violated by specific proteins. Various extensions to the algorithm are discussed.     

Topology prediction of membrane proteins.

A new method is described for prediction of protein membrane topology (intra- and extracellular sidedness) from multiply aligned amino acid sequences after determination of the membrane-spanning segments. The prediction technique relies on residue compositional differences in the protein segments exposed at each side of the membrane. Intra/extracellular ratios are calculated for the residue types Asn, Asp, Gly, Phe, Pro, Trp, Tyr, and Val, preferably found on the extracellular side, and for Ala, Arg, Cys, and Lys, mostly occurring on the intracellular side. The consensus over these 12 residue distributions is used for sidedness prediction. The method was developed with a test set of 42 protein families, for which all but one were correctly predicted with the new algorithm. This represents an improvement over predictions based on the widely used "positive-inside rule" and other techniques, where at least six mispredictions were observed for the same data set. Further, application of this and other methods to 12 protein families not in the test set still showed the better performance of the present technique, which was subsequently applied to another set of membrane protein families where the topology has yet to be determined.