Structure optimization of an artificial neural filter detecting membrane-spanning amino acid sequences

An artificial neural network has been developed for the recognition and prediction of transmembrane regions in the amino acid sequences of human integral membrane proteins. It provides an additional prediction method besides the common hydrophobicity analysis by statistical means. Membrane/nonmembrane transition regions are predicted with 92% accuracy in both training and independent test data. The method used for the development of the neural filter is the algorithm of structure evolution. It subjects both the architecture and parameters of the system to a systematical optimization process and carries out local search in the respective structure and parameter spaces. The training technique of incomplete induction as part of the structure evolution provides for a comparatively general solution of the problem that is described by input-output relations only. Seven physicochemical side-chain properties were used to encode the amino acid sequences. It was found that geometric parameters like side-chain volume, bulkiness, or surface area are of minor importance. The properties polarity, refractivity, and hydrophobicity, however, turned out to support feature extraction. It is concluded that membrane transition regions in proteins are encoded in sequences as a characteristic feature based on the respective side-chain properties. The method of structure evolution is described in detail for this particular application and suggestions for further development of amino acid sequence filters are made. © 1996 John Wiley & Sons, Inc.     

Analysis of protein transmembrane helical regions by a neural network.

Neural networks were used to generalize common themes found in transmembrane-spanning protein helices. Various-sized databases were used containing nonoverlapping sequences, each 25 amino acids long. Training consisted of sorting these sequences into 1 of 2 groups: transmembrane helical peptides or nontransmembrane peptides. Learning was measured using a test set 10% the size of the training set. As training set size increased from 214 sequences to 1,751 sequences, learning increased in a nonlinear manner from 75% to a high of 98%, then declined to a low of 87%. The final training database consisted of roughly equal numbers of transmembrane (928) and nontransmembrane (1,018) sequences. All transmembrane sequences were entered into the database with respect to their lipid membrane orientation: from inside the membrane to outside. Generalized transmembrane helix and nontransmembrane peptides were constructed from the maximally weighted connecting strengths of fully trained networks. Four generalized transmembrane helices were found to contain 9 consensus residues: a K-R-F triplet was found at the inside lipid interface, 2 isoleucine and 2 other phenylalanine residues were present in the helical body, and 2 tryptophan residues were found near the outside lipid interface. As a test of the training method, bacteriorhodopsin was examined to determine the position of its 7 transmembrane helices.     

Sequence context and modified hydrophobic moment plots help identify 'horizontal' surface helices in transmembrane protein structure prediction

Transmembrane proteins make up at least one-fifth of the genome of most organisms and are critical components of key pathways for cell survival and interactions with the environment. The function of helices found at the membrane surface in transmembrane proteins has not been greatly explored, but it is likely that they play an ancillary role to membrane spanning helices and are analogous to the surface active helices of peripheral membrane proteins, being involved in: lipid association, membrane perturbation, transmembrane signal transduction and regulation, and transmembrane helical bundle formation. Due to the difficulties in obtaining high-resolution structural data for this class of proteins, structure-from-sequence predictive methods continue to be developed as a means to obtain structural models for these largely intractable systems. A simple but effective variant of the hydrophobic moment analysis of amino acid sequences is described here as part of a protocol for distinguishing helical sequences that are parallel to or 'horizontal' at the membrane bilayer/aqueous phase interface from helices that are membrane-embedded or located in extra-membranous domains. This protocol when tested on transmembrane spanning protein amino acid sequences not used in its development, was found to be 84-91% accurate when the results were compared to the partition locations in the corresponding structures determined by X-ray crystallography, and 72% accurate in determining which helices lie horizontal or near horizontal at the lipid interface.     

Sequence similarity as a predictor of the transmembrane topology of membrane-intrinsic subunits of bacterial respiratory chain enzymes

Integral membrane proteins usually have a predominantly alpha-helical secondary structure in which transmembrane segments are connected by membrane-extrinsic loops. Although a number of membrane protein structures have been reported in recent years, in most cases transmembrane topologies are initially predicted using a variety of theoretical techniques, including hydropathy analyses and the "positive inside" rule. We have explored the use of plots of the distribution of sequence similarity within families of membrane proteins comprising homeomorphic domains as a new method for the prediction/verification of the orientation of transmembrane topology models within certain families of multimeric respiratory chain enzymes. Within such proteins, analyses of sequence similarity can: i) identify heme and/or quinol binding sites; ii) identify potential electron-transfer conduits to/from prosthetic groups; and iii) locate regions defining potential subunit-subunit interactions. We mined emerging bioinformatic data for sequences of 11 families of membrane-intrinsic proteins that are part of multimeric respiratory chain complexes that also have membrane-extrinsic subunits. The sequences of each family were then aligned and the resultant alignments converted into a graphical format recording an empirical measure of the sequence similarity plotted versus residue position. In each case, this plot was compared to the predicted transmembrane topology. With one exception, there is a strong correlation between the existence     

Understanding peptide interactions with the lipid bilayer: a guide to membrane protein engineering

In recent years, studies of the interactions of designed hydrophobic and amphipathic polypeptides with biological membranes have progressed considerably. In-plane and transmembrane helical domains have been engineered as well as sequences that exhibit dynamic distributions of different topologies. These sequences not only help our understanding of the thermodynamic interaction contributions that determine peptide orientation, but also exhibit interesting biological functions as autonomous units. In addition, helices are considered to be folding intermediates of multi-spanning membrane proteins. The specificity and thermodynamic stability of transmembrane helix-helix interactions or the assembly of beta sheets at the membrane surface have, therefore, become a focus of ongoing research.     

A solid-state NMR study of changes in lipid phase induced by membrane-fusogenic LV-peptides

Membrane fusion requires restructuring of lipid bilayers mediated by fusogenic membrane proteins. Peptides that correspond to natural transmembrane sequences or that have been designed to mimic them, such as low-complexity “Leu-Val” (LV) peptide sequences, can drive membrane fusion, presumably by disturbing the lipid bilayer structure. Here, we assess how peptides of different fusogenicity affect membrane structure using solid state NMR techniques. We find that the more fusogenic variants induce an unaligned lipid phase component and a large degree of phase separation as observed in ³¹P 2D spectra. The data support the idea that fusogenic peptides accumulate PE in a non-bilayer phase which may be critical for the induction of fusion.     

基于小波分析的膜蛋白跨膜区段序列分析和预测

膜蛋白是一类结构独特的蛋白质,在各种细胞中普遍存在,发挥着重要的生理功能。目前仅有少数膜蛋白听结构被实验测出,因此用计算机预测膜蛋白的结构是蛋白质结构预测的主要研究内容之一。膜蛋白一般在膜上形成保守的跨膜螺旋结构,序列特征明显,比较适合用预测的方法确定跨膜螺旋区段的位置。国际上已有一些研究者用人工神经网络方法、多序列比对方法和统计方法进行了预测尝试,取得了一定的成功经验。我们对蛋白质序列数据库中的     

Role of membrane potential in protein folding and domain formation during secretion in Escherichia coli

The synthesis and processing of the periplasmic components of the leucine transport system of E coli have been studied to determine the role played by transmembrane potential in protein secretion. Both the leucine-isoleucine-valine binding protein and the leucine-specific binding protein are synthesized as precursors with 23 amino acid N-terminal leader sequences. The processing of these precursors is sensitive to the transmembrane potential. Since the amino acid sequence and the crystal structure have been determined for the leucine-isoleucine-valine binding protein, it and the closely related leucine-specific binding protein represent convenient models in which to examine the mechanism of protein secretion in E coli. A model for secretion has been proposed, suggesting a role for transmembrane potential. In this model, the N-terminal amino acid sequence of the precursor is assumed to form a hairpin of two helices. The membrane potential may orient this structure to make it accessible to processing. In addition, the model suggests that a negatively charged, folded domain of the secretory protein may electrophorese toward the trans-positive side of the membrane, thus providing an additional role for the transmembrane potential.     

Prediction of the transmembrane regions of beta-barrel membrane proteins with a neural network-based predictor

A method based on neural networks is trained and tested on a nonredundant set of beta-barrel membrane proteins known at atomic resolution with a jackknife procedure. The method predicts the topography of transmembrane beta strands with residue accuracy as high as 78% when evolutionary information is used as input to the network. Of the transmembrane beta-strands included in the training set, 93% are correctly assigned. The predictor includes an algorithm of model optimization, based on dynamic programming, that correctly models eight out of the 11 proteins present in the training/testing set. In addition, protein topology is assigned on the basis of the location of the longest loops in the models. We propose this as a general method to fill the gap of the prediction of beta-barrel membrane proteins.     

Predicting transmembrane beta-barrels and interstrand residue interactions from sequence

Transmembrane beta-barrel (TMB) proteins are embedded in the outer membrane of Gram-negative bacteria, mitochondria, and chloroplasts. The cellular location and functional diversity of beta-barrel outer membrane proteins (omps) makes them an important protein class. At the present time, very few nonhomologous TMB structures have been determined by X-ray diffraction because of the experimental difficulty encountered in crystallizing transmembrane proteins. A novel method using pairwise interstrand residue statistical potentials derived from globular (nonouter membrane) proteins is introduced to predict the supersecondary structure of transmembrane beta-barrel proteins. The algorithm transFold employs a generalized hidden Markov model (i.e., multitape S-attribute grammar) to describe potential beta-barrel supersecondary structures and then computes by dynamic programming the minimum free energy beta-barrel structure. Hence, the approach can be viewed as a "wrapping" component that may capture folding processes with an initiation stage followed by progressive interaction of the sequence with the already-formed motifs. This approach differs significantly from others, which use traditional machine learning to solve this problem, because it does not require a training phase on known TMB structures and is the first to explicitly capture and predict long-range interactions. TransFold outperforms previous programs for predicting TMBs on smaller (相似文献   

Membrane insertion scanning of the human ileal sodium/bile acid co-transporter.

Mammalian sodium-dependent bile acid transporters (SBATs) responsible for bile salt uptake across the liver sinusoidal or ileal/renal brush border membrane have been identified and share approximately 35% amino acid sequence identity. Programs for prediction of topology and localization of transmembrane helices identify eight or nine hydrophobic regions for the SBAT sequences as membrane spanning. Analysis of N-linked glycosylation has provided evidence for an exoplasmic N-terminus and a cytoplasmic C-terminus, indicative of an odd number of transmembrane segments. To determine the membrane topography of the human ileal SBAT (HISBAT), an in vitro translation/translocation protocol was employed using three different fusion protein constructs. Individual HISBAT segments were analyzed for signal anchor or stop translocation (stop transfer) activity by insertion between a cytoplasmic anchor (HK M0) or a signal anchor segment (HK M1) and a glycosylation flag (HK beta). To examine consecutive HISBAT sequences, sequential hydrophobic sequences were inserted into the HK M0 vector or fusion vectors were made that included the glycosylated N-terminus of HISBAT, sequential hydrophobic sequences, and the glycosylation flag. Individual signal anchor (SA) and stop transfer (ST) properties were found for seven out of the nine predicted hydrophobic segments (H1, H2, H4, H5, H6, H7, and H9), supporting a seven transmembrane segment model. However, the H3 region was membrane inserted when translated in the context of the native HISBAT flanking sequences. Furthermore, results from translations of sequential constructs ending after H7 provided support for integration of H8. These data provide support for a SBAT transmembrane domain model with nine integrated segments with an exoplasmic N-terminus and a cytoplasmic C-terminus consistent with a recent predictive analysis of this transporter topology.     

How Many 3D Structures Do We Need to Train a Predictor？

It has been shown that the progress in the determination of membrane protein structure grows exponentially, with approximately the same growth rate as that of the water-soluble proteins. In order to investigate the effect of this, on the performance of prediction algorithms for both α-helical and β-barrel membrane proteins, we conducted a prospective study based on historical records. We trained separate hidden Markov models with different sized training sets and evaluated their performance on topology pred...     

A Simple Method for Predicting Transmembrane Proteins Based on Wavelet Transform

The increasing protein sequences from the genome project require theoretical methods to predict transmembrane helical segments (TMHs). So far, several prediction methods have been reported, but there are some deficiencies in prediction accuracy and adaptability in these methods. In this paper, a method based on discrete wavelet transform (DWT) has been developed to predict the number and location of TMHs in membrane proteins. PDB coded as 1KQG is chosen as an example to describe the prediction process by this method. 80 proteins with known 3D structure from Mptopo database are chosen at random as data sets (including 325 TMHs) and 80 sequences are divided into 13 groups according to their function and type. TMHs prediction is carried out for each group of membrane protein sequences and obtain satisfactory result. To verify the feasibility of this method, 80 membrane protein sequences are treated as test sets, 308 TMHs can be predicted and the prediction accuracy is 96.3%. Compared with the main prediction results of seven popular prediction methods, the obtained results indicate that the proposed method in this paper has higher prediction accuracy.     

Photoinduced reactivity of the HIV-1 envelope glycoprotein with a membrane-embedded probe reveals insertion of portions of the HIV-1 Gp41 cytoplasmic tail into the viral membrane

The interactions of HIV-1 Env (gp120-gp41) with CD4 and coreceptors trigger a barrage of conformational changes in Env that drive the membrane fusion process. Various regions of gp41 have profound effects on HIV entry and budding. However, the precise interactions between gp41 and the membrane have not been elucidated. To examine portions of membrane proteins that are embedded in membrane lipids, we have studied photoinduced chemical reactions in membranes using the lipid bilayer specific probe iodonaphthyl azide (INA). Here we show that in addition to the transmembrane anchor, amphipatic sequences in the cytoplasmic tail (CT) of HIV-1 gp41 are labeled by INA. INA labeling of the HIV-1 gp41 CT was similar whether wild-type or a mutant HIV-1 was used with uncleaved p55 Gag, which does not allow entry. These results shed light on the disposition of the HIV-1 gp41 CT with respect to the membrane. Moreover, our data have general implications for topology of membrane proteins and their in situ interactions with the lipid bilayer.     

Towards a comparative anatomy of N-terminal topogenic protein sequences

A comparative study of three kinds of eukaryotic N-terminal topogenic sequences, viz signal peptides, N-terminal transmembrane anchors, and mitochondrial targeting sequences, suggests: that the sign of the N-terminal charge might influence the orientation of an N-terminal hydrophobic segment relative to the membrane and give rise to N-terminally anchored proteins with their main mass exposed either on the cytosolic or extra-cytosolic side of the membrane; and that N-terminal transmembrane segments in mitochondrial targeting sequences have a relatively low overall hydrophobicity, probably in order to avoid being recognized by the endoplasmic reticulum export machinery.     

The structure of the mouse immunoglobulin in gamma 3 membrane gene segment

The genomic region containing the mouse immunoglobulin gamma 3 heavy chain membrane (M) exons has been located and sequenced. The exon structure is highly similar to that of the other mouse gamma chains, with strong sequence conservation in the coding regions and the intron 5' to the M1 exon. The intron between M1 and M2 shows moderate sequence homology but very strong conservation of size. RNA blots suggest that gamma 3 membrane exon usage is similar to that seen in other immunoglobulin membrane heavy chain mRNAs. The transmembrane region contains the invariant residues which have been noted in all other heavy chain sequences and which were previously proposed to be interactive in a two-chain model for insertion through the lymphocyte membrane. Conserved residues with similar spacing have been seen in class II histocompatibility antigens, which are also two-chain transmembrane molecules, but not in class I antigens, which span cell membranes with a single chain.     

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.     

Definition of the nuclear encoded protein composition of bovine heart mitochondrial complex I. Identification of two new subunits

Mitochondrial NADH:ubiquinone oxidoreductase (complex I) from bovine heart is a complicated multisubunit, membrane-bound assembly. Seven subunits are encoded by mitochondrial DNA, and the sequences of 36 nuclear encoded subunits have been described. The subunits of complex I and two subcomplexes (Ialpha and Ibeta) were resolved on one- and two-dimensional gels and by reverse-phase high performance liquid chromatography. Mass spectrometric analysis revealed two previously unknown subunits in complex I, named B14.7 and ESSS, one in each subcomplex. Coding sequences for each protein were identified in data bases and were confirmed by cDNA cloning and sequencing. Subunit B14.7 has an acetylated N terminus, no presequence, and contains four potential transmembrane helices. It is homologous to subunit 21.3b from complex I in Neurospora crassa and is related to Tim17, Tim22, and Tim23, which are involved in protein translocation across the inner membrane. Subunit ESSS has a cleaved mitochondrial import sequence and one potential transmembrane helix. A total of 45 different subunits of bovine complex I have now been characterized.     

Interaction of influenza virus haemagglutinin with sphingolipid-cholesterol membrane domains via its transmembrane domain.

Sphingolipid-cholesterol rafts are microdomains in biological membranes with liquid-ordered phase properties which are implicated in membrane traffic and signalling events. We have used influenza virus haemagglutinin (HA) as a model protein to analyse the interaction of transmembrane proteins with these microdomains. Here we demonstrate that raft association is an intrinsic property encoded in the protein. Mutant HA molecules with foreign transmembrane domain (TMD) sequences lose their ability to associate with the lipid microdomains, and mutations in the HA TMD reveal a requirement for hydrophobic residues in contact with the exoplasmic leaflet of the membrane. We also provide experimental evidence that cholesterol is critically required for association of proteins with lipid rafts. Our data suggest that the binding to specific membrane domains can be encoded in transmembrane proteins and that this information will be used for polarized sorting and signal transduction processes.