Structural bioinformatics prediction of membrane-binding proteins

Interfacial Enzymes Enable Gram-Positive Microbes to Eat Fatty Acids

Exogenous fatty acid (eFA) activation and utilization play key roles in bacterial physiology and confer growth advantages by bypassing the need to make fatty acids for lipid synthesis. In Gram-positive bacteria, eFA activation and utilization is generally carried out by the fatty acid kinase (FakAB) two-component system that converts eFA to acyl phosphate, and the acyl-ACP:phosphate transacylase (PlsX) that catalyzes the reversible conversion of acyl phosphate to acyl-acyl carrier protein. Acyl-acyl carrier protein is a soluble format of the fatty acid that is compatible with cellular metabolic enzymes and can feed multiple processes including the fatty acid biosynthesis pathway. The combination of FakAB and PlsX enables the bacteria to channel eFA nutrients. These key enzymes are peripheral membrane interfacial proteins that associate with the membrane through amphipathic helices and hydrophobic loops. In this review, we discuss the biochemical and biophysical advances that have established the structural features that drive FakB or PlsX association with the membrane, and how these protein-lipid interactions contribute to enzyme catalysis.

DREAMM: a web-based server for drugging protein-membrane interfaces as a novel workflow for targeted drug design

SUMMARY

The allosteric modulation of peripheral membrane proteins (PMPs) by targeting protein-membrane interactions with drug-like molecules represents a new promising therapeutic strategy for proteins currently considered undruggable. However, the accessibility of protein-membrane interfaces by small molecules has been so far unexplored, possibly due to the complexity of the interface, the limited protein-membrane structural information and the lack of computational workflows to study it. Herein, we present a pipeline for drugging protein-membrane interfaces using the DREAMM (Drugging pRotein mEmbrAne Machine learning Method) web server. DREAMM works in the back end with a fast and robust ensemble machine learning algorithm for identifying protein-membrane interfaces of PMPs. Additionally, DREAMM also identifies binding pockets in the vicinity of the predicted membrane-penetrating amino acids in protein conformational ensembles provided by the user or generated within DREAMM.

AVAILABILITY AND IMPLEMENTATION

DREAMM web server is accessible via https://dreamm.ni4os.eu.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Predicting protein–membrane interfaces of peripheral membrane proteins using ensemble machine learning

Abnormal protein–membrane attachment is involved in deregulated cellular pathways and in disease. Therefore, the possibility to modulate protein–membrane interactions represents a new promising therapeutic strategy for peripheral membrane proteins that have been considered so far undruggable. A major obstacle in this drug design strategy is that the membrane-binding domains of peripheral membrane proteins are usually unknown. The development of fast and efficient algorithms predicting the protein–membrane interface would shed light into the accessibility of membrane–protein interfaces by drug-like molecules. Herein, we describe an ensemble machine learning methodology and algorithm for predicting membrane-penetrating amino acids. We utilize available experimental data from the literature for training 21 machine learning classifiers and meta-classifiers. Evaluation of the best ensemble classifier model accuracy yields a macro-averaged F1 score = 0.92 and a Matthews correlation coefficient = 0.84 for predicting correctly membrane-penetrating amino acids on unknown proteins of a validation set. The python code for predicting protein–membrane interfaces of peripheral membrane proteins is available at https://github.com/zoecournia/DREAMM.

Computational Approaches to Investigate and Design Lipid-binding Domains for Membrane Biosensing

Association of proteins with cellular membranes is critical for signaling and membrane trafficking processes. Many peripheral lipid-binding domains have been identified in the last few decades and have been investigated for their specific lipid-sensing properties using traditional in vivo and in vitro studies. However, several knowledge gaps remain owing to intrinsic limitations of these methodologies. Thus, novel approaches are necessary to further our understanding in lipid-protein biology. This review briefly discusses lipid-binding domains that act as specific lipid biosensors and provides a broad perspective on the computational approaches such as molecular dynamics (MD) simulations and machine learning (ML)-based techniques that can be used to study protein-membrane interactions. We also highlight the need for de novo design of proteins that elicit specific lipid-binding properties.

PerMemDB: A database for eukaryotic peripheral membrane proteins

The majority of all proteins in cells interact with membranes either permanently or temporarily. Peripheral membrane proteins form transient complexes with membrane proteins and/or lipids, via non-covalent interactions and are of outmost importance, due to numerous cellular functions in which they participate. In an effort to collect data regarding this heterogeneous group of proteins we designed and constructed a database, called PerMemDB. PerMemDB is currently the most complete and comprehensive repository of data for eukaryotic peripheral membrane proteins deposited in UniProt or predicted with the use of MBPpred - a computational method that specializes in the detection of proteins that interact non-covalently with membrane lipids, via membrane binding domains. The first version of the database contains 231,770 peripheral membrane proteins from 1009 organisms. All entries have cross-references to other databases, literature references and annotation regarding their interactions with other proteins. Moreover, additional sequence annotation of the characteristic domains that allow these proteins to interact with membranes is available, due to the application of MBPpred. Through the web interface of PerMemDB, users can browse the contents of the database, submit advanced text searches and BLAST queries against the protein sequences deposited in PerMemDB. We expect this repository to serve as a source of information that will allow the scientific community to gain a deeper understanding of the evolution and function of peripheral membrane proteins via the enhancement of proteome-wide analyses. The database is available at: http://bioinformatics.biol.uoa.gr/db=permemdb.

A High-Throughput Fluorometric Assay for Lipid-Protein Binding

An increasing number of intracellular and extracellular proteins are shown to interact with membrane lipids under physiological conditions. For rapid and robust quantitative measurement of lipid-protein interaction, we developed a sensitive fluorescence quenching-based assay that is universally applicable to all proteins and lipids. The assay employs fluorescence protein (FP)-tagged proteins whose fluorescence emission intensity is decreased when they bind vesicles containing quenching lipids. This simple assay can be performed with a fluorescence plate reader or a spectrofluorometer and optimized for different proteins with various combinations of FPs and quenching lipids. The assay allows a rapid, sensitive, and accurate determination of lipid specificity and affinity for various lipid-binding proteins, and high-throughput screening of molecules that modulate their membrane binding.

Bioavailability of Fullerene under Environmentally Relevant Conditions: Effects of Humic Acid and Fetal Bovine Serum on Accumulation in Lipid Bilayers and Cellular Uptake

Carbon fullerene (C60) has emerged at the forefront of nanoscale research and application due to its unique properties. As the production of this nanoparticle rapidly increases, it can be released into natural aquatic environments and can accumulate in biological systems. This research examined the effects of humic acid and fetal bovine serum (FBS), which are ubiquitous in aquatic environments and representative of blood plasma in living organisms, respectively, on bioavailability of fullerene. Bioavailability was investigated using in vitro methods for lipid membrane accumulation and cellular uptake studies. Humic acid and FBS significantly changed the characteristics of fullerene including its particle size and surface charge. The effects of humic acid on lipid accumulation of fullerene depended on the lipid head charge. FBS also significantly decreased the lipid accumulation when positively charged and zwitterionic head groups were present on the lipids, possibly due to the higher steric repulsion of the protein coated nanoparticles. In addition, both humic acid and FBS protein effectively lowered the amounts of fullerene taken up by Caco-2 cells, which are derived from a human colorectal adenocarcinoma and have similar functions to the small intestinal epithelium. Results of this study suggest that surface modification of fullerene by environmentally relevant matrices can significantly affect the biological transport, as well as the possible toxicity of this nanomaterial.

High-Throughput Fluorometric Assay for Membrane-Protein Interaction

Membrane-protein interaction plays key roles in a wide variety of biological processes. To facilitate rapid and sensitive measurement of membrane binding of soluble proteins, we developed a fluorescence-based quantitative assay that is universally applicable to all proteins. This fluorescence-quenching assay employs fluorescence protein (FP)-tagged proteins whose fluorescence intensity is greatly decreased when they bind vesicles containing synthetic lipid dark quenchers, such as N-dimethylaminoazobenzenesulfonylphosphatidylethanolamine (dabsyl-PE). This simple assay can be performed with either a spectrofluorometer or a plate reader and optimized for different proteins with various combinations of FPs and quenching lipids. The assay allows rapid, sensitive, and accurate determination of lipid specificity and affinity for various lipid binding domains and proteins, and also high-throughput screening of small molecules that modulate membrane binding of proteins.

Transmembrane proteins--Mining the cattle tick transcriptome

Managing the spread and load of pathogen-transmitting ticks is an important task worldwide. The cattle tick, Rhipicephalus microplus, not only impacts the economy through losses in dairy and meat production, but also raises concerns for human health in regards to the potential of certain transmitted pathogens becoming zoonotic. However, novel strategies to control R. microplus are hindered by lack of understanding tick biology and the discovery of suitable vaccine or acaricide targets. The importance of transmembrane proteins as vaccine targets are well known, as is the case in tick vaccines with Bm86 as antigen. In this study, we describe the localization and functional annotation of 878 putative transmembrane proteins. Thirty proteins could be confirmed in the R. microplus gut using LC-MS/MS analysis and their roles in tick biology are discussed. To the best of our knowledge, 19 targets have not been reported before in any proteomics study in various tick species and the possibility of using the identified proteins as targets for tick control are discussed. Although tissue expression of identified putative proteins through expansive proteomics is necessary, this study demonstrates the possibility of using bioinformatics for the identification of targets for further evaluation in tick control strategies.

Neural network and SVM classifiers accurately predict lipid binding proteins, irrespective of sequence homology

Due to the central roles of lipid binding proteins (LBPs) in many biological processes, sequence based identification of LBPs is of great interest. The major challenge is that LBPs are diverse in sequence, structure, and function which results in low accuracy of sequence homology based methods. Therefore, there is a need for developing alternative functional prediction methods irrespective of sequence similarity. To identify LBPs from non-LBPs, the performances of support vector machine (SVM) and neural network were compared in this study. Comprehensive protein features and various techniques were employed to create datasets. Five-fold cross-validation (CV) and independent evaluation (IE) tests were used to assess the validity of the two methods. The results indicated that SVM outperforms neural network. SVM achieved 89.28% (CV) and 89.55% (IE) overall accuracy in identification of LBPs from non-LBPs and 92.06% (CV) and 92.90% (IE) (in average) for classification of different LBPs classes. Increasing the number and the range of extracted protein features as well as optimization of the SVM parameters significantly increased the efficiency of LBPs class prediction in comparison to the only previous report in this field. Altogether, the results showed that the SVM algorithm can be run on broad, computationally calculated protein features and offers a promising tool in detection of LBPs classes. The proposed approach has the potential to integrate and improve the common sequence alignment based methods.

Cellular and molecular interactions of phosphoinositides and peripheral proteins

Anionic lipids act as signals for the recruitment of proteins containing cationic clusters to biological membranes. A family of anionic lipids known as the phosphoinositides (PIPs) are low in abundance, yet play a critical role in recruitment of peripheral proteins to the membrane interface. PIPs are mono-, bis-, or trisphosphorylated derivatives of phosphatidylinositol (PI) yielding seven species with different structure and anionic charge. The differential spatial distribution and temporal appearance of PIPs is key to their role in communicating information to target proteins. Selective recognition of PIPs came into play with the discovery that the substrate of protein kinase C termed pleckstrin possessed the first PIP binding region termed the pleckstrin homology (PH) domain. Since the discovery of the PH domain, more than ten PIP binding domains have been identified including PH, ENTH, FYVE, PX, and C2 domains. Representative examples of each of these domains have been thoroughly characterized to understand how they coordinate PIP headgroups in membranes, translocate to specific membrane docking sites in the cell, and function to regulate the activity of their full-length proteins. In addition, a number of novel mechanisms of PIP-mediated membrane association have emerged, such as coincidence detection-specificity for two distinct lipid headgroups. Other PIP-binding domains may also harbor selectivity for a membrane physical property such as charge or membrane curvature. This review summarizes the current understanding of the cellular distribution of PIPs and their molecular interaction with peripheral proteins.

Global approaches for the elucidation of phosphoinositide-binding proteins

Phosphoinositide lipids (PIPns) control numerous critical biological pathways, typically through the regulation of protein function driven by non-covalent protein-lipid binding interactions. Despite the importance of these systems, the unraveling of the full scope of protein-PIPn interactions has represented a significant challenge due to the massive complexity associated with these events, including the large number of diverse proteins that bind to these lipids, variations in the mechanisms by which proteins bind to lipids, and the presence of multiple distinct PIPn isomers. As a result of this complexity, global methods in which numerous proteins that bind PIPns can be identified and characterized simultaneously from complex samples, which have been enabled by key technological advancements, have become popular as an efficient means for tackling this challenge. This review article provides an overview of advancements in large-scale methods for profiling protein-PIPn binding, including experimental methods, such as affinity enrichment, microarray analysis and activity-based protein profiling, as well as computational methods, and combined computational/experimental efforts.

High-throughput fluorescence assay for membrane-protein interaction

Membrane-protein interaction plays key roles in a wide variety of biological processes. Although various methods have been employed to measure membrane binding of soluble proteins, a robust high-throughput assay that is universally applicable to all proteins is lacking at present. Here we report a new fluorescence quenching assay utilizing enhanced green fluorescence protein (EGFP)-fusion proteins and a lipid containing a dark quencher, N-dimethylaminoazobenzenesulfonyl-phosphatidylethanolamine (dabsyl-PE). The EGFP fluorescence emission intensity showed a large decrease (i.e., >50%) when EGFP-fusion proteins bound the vesicles containing 5 mol% dabsyl-PE. This simple assay, which can be performed using either a cuvette-based spectrofluorometer or a fluorescence plate reader, allowed rapid, sensitive, and accurate determination of lipid specificity and affinity for various lipid binding domains, including two pleckstrin homology domains, an epsin N-terminal homology domain, and a phox homology domain. The assay can also be applied to high-throughput screening of small molecules that modulate membrane binding of proteins.

Calciomics: integrative studies of Ca2+-binding proteins and their interactomes in biological systems

Calcium ion (Ca(2+)), the fifth most common chemical element in the earth's crust, represents the most abundant mineral in the human body. By binding to a myriad of proteins distributed in different cellular organelles, Ca(2+) impacts nearly every aspect of cellular life. In prokaryotes, Ca(2+) plays an important role in bacterial movement, chemotaxis, survival reactions and sporulation. In eukaryotes, Ca(2+) has been chosen through evolution to function as a universal and versatile intracellular signal. Viruses, as obligate intracellular parasites, also develop smart strategies to manipulate the host Ca(2+) signaling machinery to benefit their own life cycles. This review focuses on recent advances in applying both bioinformatic and experimental approaches to predict and validate Ca(2+)-binding proteins and their interactomes in biological systems on a genome-wide scale (termed "calciomics"). Calmodulin is used as an example of Ca(2+)-binding protein (CaBP) to demonstrate the role of CaBPs on the regulation of biological functions. This review is anticipated to rekindle interest in investigating Ca(2+)-binding proteins and Ca(2+)-modulated functions at the systems level in the post-genomic era.

A structure-based protocol for learning the family-specific mechanisms of membrane-binding domains

MOTIVATION

Peripheral membrane-targeting domain (MTD) families, such as C1-, C2- and PH domains, play a key role in signal transduction and membrane trafficking by dynamically translocating their parent proteins to specific plasma membranes when changes in lipid composition occur. It is, however, difficult to determine the subset of domains within families displaying this property, as sequence motifs signifying the membrane binding properties are not well defined. For this reason, procedures based on sequence similarity alone are often insufficient in computational identification of MTDs within families (yielding less than 65% accuracy even with a sequence identity of 70%).

RESULTS

We present a machine learning protocol for determining membrane-targeting properties achieving 85-90% accuracy in separating binding and non-binding domains within families. Our model is based on features from both sequence and structure, thereby incorporation statistics obtained from the entire domain family and domain-specific physical quantities such as surface electrostatics. In addition, by using the enriched rules in alternating decision tree classifiers, we are able to determine the meaning of the assigned function labels in terms of biological mechanisms.

CONCLUSIONS

The high accuracy of the learned models and good agreement between the rules discovered using the ADtree classifier and mechanisms reported in the literature reflect the value of machine learning protocols in both prediction and biological knowledge discovery. Our protocol can thus potentially be used as a general function annotation and knowledge mining tool for other protein domains.

AVAILABILITY

metador.bioengr.uic.edu

CONTACT

huilu@uic.edu.

Predicting membrane protein types by incorporating protein topology, domains, signal peptides, and physicochemical properties into the general form of Chou's pseudo amino acid composition

The type information of un-annotated membrane proteins provides an important hint for their biological functions. The experimental determination of membrane protein types, despite being more accurate and reliable, is not always feasible due to the costly laboratory procedures, thereby creating a need for the development of bioinformatics methods. This article describes a novel computational classifier for the prediction of membrane protein types using proteins' sequences. The classifier, comprising a collection of one-versus-one support vector machines, makes use of the following sequence attributes: (1) the cationic patch sizes, the orientation, and the topology of transmembrane segments; (2) the amino acid physicochemical properties; (3) the presence of signal peptides or anchors; and (4) the specific protein motifs. A new voting scheme was implemented to cope with the multi-class prediction. Both the training and the testing sequences were collected from SwissProt. Homologous proteins were removed such that there is no pair of sequences left in the datasets with a sequence identity higher than 40%. The performance of the classifier was evaluated by a Jackknife cross-validation and an independent testing experiments. Results show that the proposed classifier outperforms earlier predictors in prediction accuracy in seven of the eight membrane protein types. The overall accuracy was increased from 78.3% to 88.2%. Unlike earlier approaches which largely depend on position-specific substitution matrices and amino acid compositions, most of the sequence attributes implemented in the proposed classifier have supported literature evidences. The classifier has been deployed as a web server and can be accessed at http://bsaltools.ym.edu.tw/predmpt.

Genome-wide functional annotation of dual-specificity protein- and lipid-binding modules that regulate protein interactions

Emerging evidence indicates that membrane lipids regulate protein networking by directly interacting with protein-interaction domains (PIDs). As a pilot study to identify and functionally annodate lipid-binding PIDs on a genomic scale, we performed experimental and computational studies of PDZ domains. Characterization of 70 PDZ domains showed that ~40% had submicromolar membrane affinity. Using a computational model built from these data, we predicted the membrane-binding properties of 2,000 PDZ domains from 20 species. The accuracy of the prediction was experimentally validated for 26 PDZ domains. We also subdivided lipid-binding PDZ domains into three classes based on the interplay between membrane- and protein-binding sites. For different classes of PDZ domains, lipid binding regulates their protein interactions by different mechanisms. Functional studies of a PDZ domain protein, rhophilin 2, suggest that all classes of lipid-binding PDZ domains serve as genuine dual-specificity modules regulating protein interactions at the membrane under physiological conditions.

Emerging methodologies to investigate lipid-protein interactions

Cellular membranes are composed of hundreds of different lipids, ion channels, receptors and scaffolding complexes that act as signalling and trafficking platforms for processes fundamental to life. Cellular signalling and membrane trafficking are often regulated by peripheral proteins, which reversibly interact with lipid molecules in highly regulated spatial and temporal fashions. In most cases, one or more modular lipid-binding domain(s) mediate recruitment of peripheral proteins to specific cellular membranes. These domains, of which more than 10 have been identified since 1989, harbour structurally selective lipid-binding sites. Traditional in vitro and in vivo studies have elucidated how these domains coordinate their cognate lipids and thus how the parent proteins associate with membranes. Cellular activities of peripheral proteins and subsequent physiological processes depend upon lipid binding affinities and selectivity. Thus, the development of novel sensitive and quantitative tools is essential in furthering our understanding of the function and regulation of these proteins. As this field expands into new areas such as computational biology, cellular lipid mapping, single molecule imaging, and lipidomics, there is an urgent need to integrate technologies to detail the molecular architecture and mechanisms of lipid signalling. This review surveys emerging cellular and in vitro approaches for studying protein-lipid interactions and provides perspective on how integration of methodologies directs the future development of the field.

Biophysics of α-synuclein membrane interactions

Membrane proteins participate in nearly all cellular processes; however, because of experimental limitations, their characterization lags far behind that of soluble proteins. Peripheral membrane proteins are particularly challenging to study because of their inherent propensity to adopt multiple and/or transient conformations in solution and upon membrane association. In this review, we summarize useful biophysical techniques for the study of peripheral membrane proteins and their application in the characterization of the membrane interactions of the natively unfolded and Parkinson's disease (PD) related protein, α-synuclein (α-syn). We give particular focus to studies that have led to the current understanding of membrane-bound α-syn structure and the elucidation of specific membrane properties that affect α-syn-membrane binding. Finally, we discuss biophysical evidence supporting a key role for membranes and α-syn in PD pathogenesis. This article is part of a Special Issue entitled: Membrane protein structure and function.

Genome-wide structural analysis reveals novel membrane binding properties of AP180 N-terminal homology (ANTH) domains

An increasing number of cytosolic proteins are shown to interact with membrane lipids during diverse cellular processes, but computational prediction of these proteins and their membrane binding behaviors remains challenging. Here, we introduce a new combinatorial computation protocol for systematic and robust functional prediction of membrane-binding proteins through high throughput homology modeling and in-depth calculation of biophysical properties. The approach was applied to the genomic scale identification of the AP180 N-terminal homology (ANTH) domain, one of the modular lipid binding domains, and prediction of their membrane binding properties. Our analysis yielded comprehensive coverage of the ANTH domain family and allowed classification and functional annotation of proteins based on the differences in local structural and biophysical features. Our analysis also identified a group of plant ANTH domains with unique structural features that may confer novel functionalities. Experimental characterization of a representative member of this subfamily confirmed its unique membrane binding mechanism and unprecedented membrane deforming activity. Collectively, these studies suggest that our new computational approach can be applied to genome-wide functional prediction of other lipid binding domains.

Tryptophan residues promote membrane association for a plant lipid glycosyltransferase involved in phosphate stress

Chloroplast membranes contain a substantial excess of the nonbilayer-prone monogalactosyldiacylglycerol (GalDAG) over the biosynthetically consecutive, bilayer-forming digalactosyldiacylglycerol (GalGalDAG), yielding a high membrane curvature stress. During phosphate shortage, plants replace phospholipids with GalGalDAG to rescue phosphate while maintaining membrane homeostasis. Here we investigate how the activity of the corresponding glycosyltransferase (GT) in Arabidopsis thaliana (atDGD2) depends on local bilayer properties by analyzing structural and activity features of recombinant protein. Fold recognition and sequence analyses revealed a two-domain GT-B monotopic structure, present in other plant and bacterial glycolipid GTs, such as the major chloroplast GalGalDAG GT atDGD1. Modeling led to the identification of catalytically important residues in the active site of atDGD2 by site-directed mutagenesis. The DGD synthases share unique bilayer interface segments containing conserved tryptophan residues that are crucial for activity and for membrane association. More detailed localization studies and liposome binding analyses indicate differentiated anchor and substrate-binding functions for these separated enzyme interface regions. Anionic phospholipids, but not curvature-increasing nonbilayer lipids, strongly stimulate enzyme activity. From our studies, we propose a model for bilayer "control" of enzyme activity, where two tryptophan segments act as interface anchor points to keep the substrate region close to the membrane surface. Binding of the acceptor substrate is achieved by interaction of positive charges in a surface cluster of lysines, arginines, and histidines with the surrounding anionic phospholipids. The diminishing phospholipid fraction during phosphate shortage stress will then set the new GalGalDAG/phospholipid balance by decreasing stimulation of atDGD2.

An improved machine learning protocol for the identification of correct Sequest search results

Background

Mass spectrometry has become a standard method by which the proteomic profile of cell or tissue samples is characterized. To fully take advantage of tandem mass spectrometry (MS/MS) techniques in large scale protein characterization studies robust and consistent data analysis procedures are crucial. In this work we present a machine learning based protocol for the identification of correct peptide-spectrum matches from Sequest database search results, improving on previously published protocols.

Results

The developed model improves on published machine learning classification procedures by 6% as measured by the area under the ROC curve. Further, we show how the developed model can be presented as an interpretable tree of additive rules, thereby effectively removing the 'black-box' notion often associated with machine learning classifiers, allowing for comparison with expert rule-of-thumb. Finally, a method for extending the developed peptide identification protocol to give probabilistic estimates of the presence of a given protein is proposed and tested.

Conclusions

We demonstrate the construction of a high accuracy classification model for Sequest search results from MS/MS spectra obtained by using the MALDI ionization. The developed model performs well in identifying correct peptide-spectrum matches and is easily extendable to the protein identification problem. The relative ease with which additional experimental parameters can be incorporated into the classification framework, to give additional discriminatory power, allows for future tailoring of the model to take advantage of information from specific instrument set-ups.

Boosting the prediction and understanding of DNA-binding domains from sequence

DNA-binding proteins perform vital functions related to transcription, repair and replication. We have developed a new sequence-based machine learning protocol to identify DNA-binding proteins. We compare our method with an extensive benchmark of previously published structure-based machine learning methods as well as a standard sequence alignment technique, BLAST. Furthermore, we elucidate important feature interactions found in a learned model and analyze how specific rules capture general mechanisms that extend across DNA-binding motifs. This analysis is carried out using the malibu machine learning workbench available at http://proteomics.bioengr.uic.edu/malibu and the corresponding data sets and features are available at http://proteomics.bioengr.uic.edu/dna.

Genome-wide sequence-based prediction of peripheral proteins using a novel semi-supervised learning technique

BACKGROUND

In supervised learning, traditional approaches to building a classifier use two sets of examples with pre-defined classes along with a learning algorithm. The main limitation of this approach is that examples from both classes are required which might be infeasible in certain cases, especially those dealing with biological data. Such is the case for membrane-binding peripheral domains that play important roles in many biological processes, including cell signaling and membrane trafficking by reversibly binding to membranes. For these domains, a well-defined positive set is available with domains known to bind membrane along with a large unlabeled set of domains whose membrane binding affinities have not been measured. The aforementioned limitation can be addressed by a special class of semi-supervised machine learning called positive-unlabeled (PU) learning that uses a positive set with a large unlabeled set. METHODS In this study, we implement the first application of PU-learning to a protein function prediction problem: identification of peripheral domains. PU-learning starts by identifying reliable negative (RN) examples iteratively from the unlabeled set until convergence and builds a classifier using the positive and the final RN set. A data set of 232 positive cases and ~3750 unlabeled ones were used to construct and validate the protocol.

RESULTS

Holdout evaluation of the protocol on a left-out positive set showed that the accuracy of prediction reached up to 95% during two independent implementations.

CONCLUSION

These results suggest that our protocol can be used for predicting membrane-binding properties of a wide variety of modular domains. Protocols like the one presented here become particularly useful in the case of availability of information from one class only.

Using Correlated Monte Carlo Sampling for Efficiently Solving the Linearized Poisson-Boltzmann Equation Over a Broad Range of Salt Concentration

Dielectric continuum or implicit solvent models provide a significant reduction in computational cost when accounting for the salt-mediated electrostatic interactions of biomolecules immersed in an ionic environment. These models, in which the solvent and ions are replaced by a dielectric continuum, seek to capture the average statistical effects of the ionic solvent, while the solute is treated at the atomic level of detail. For decades, the solution of the three-dimensional Poisson-Boltzmann equation (PBE), which has become a standard implicit-solvent tool for assessing electrostatic effects in biomolecular systems, has been based on various deterministic numerical methods. Some deterministic PBE algorithms have drawbacks, which include a lack of properly assessing their accuracy, geometrical difficulties caused by discretization, and for some problems their cost in both memory and computation time. Our original stochastic method resolves some of these difficulties by solving the PBE using the Monte Carlo method (MCM). This new approach to the PBE is capable of efficiently solving complex, multi-domain and salt-dependent problems in biomolecular continuum electrostatics to high precision. Here we improve upon our novel stochastic approach by simultaneouly computating of electrostatic potential and solvation free energies at different ionic concentrations through correlated Monte Carlo (MC) sampling. By using carefully constructed correlated random walks in our algorithm, we can actually compute the solution to a standard system including the linearized PBE (LPBE) at all salt concentrations of interest, simultaneously. This approach not only accelerates our MCPBE algorithm, but seems to have cost and accuracy advantages over deterministic methods as well. We verify the effectiveness of this technique by applying it to two common electrostatic computations: the electrostatic potential and polar solvation free energy for calcium binding proteins that are compared with similar results obtained using mature deterministic PBE methods.

Enhanced biological stability of collagen with incorporation of PAMAM dendrimer

The crosslinking methods of collagen using glutaraldehyde (GTA) and 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide (EDC)/N-hydroxysuccinimide (NHS) are frequently performed in biomedical applications, but both methods still have their own disadvantages, including the GTA cytotoxicity and low degree of EDC/NHS crosslinking. In this study, we incorporated polyamidoamine (PAMAM) dendrimer with surface amine groups into the two aforementioned crosslinking methods to improve the biostability and structural integrity of collagen. Fifty micromolar of dendrimer concentration was found to have negligible in vitro cytotoxicity and was used for EDC and GTA crosslinking of collagen. The collagenase digestion assay showed that the collagen scaffolds crosslinked in the presence of PAMAM exhibited a higher denature temperature and higher resistance against collagenase digestion compared with their counterparts without dendrimer. Cell proliferation with human conjunctival fibroblasts showed that the incorporation of PAMAM in EDC crosslinking significantly increased the proliferation. All the crosslinked scaffolds also exhibited higher structural stability than the noncrosslinked scaffold. Crosslinking with EDC and PAMAM together yielded substantially higher proliferation and may be a suitable collagen scaffold for biomedical applications.

Effect of bilayer phospholipid composition and curvature on ligand transfer by the alpha-tocopherol transfer protein

We report here our preliminary investigations on the mechanism of alpha-TTP-mediated ligand transfer as assessed using fluorescence resonance energy transfer (FRET) assays. These assays monitor the movement of the model alpha-tocopherol fluorescent derivative ((R)-2,5,7,8-tetramethyl-chroman-2-[9-(7-nitro-benzo[1,2,5]oxadiazol-4-yl amino)-nonyl]-chroman-6-ol; NBD-Toc) from protein to acceptor vesicles containing the fluorescence quencher TRITC-PE. We have found that alpha-TTP utilizes a collisional mechanism of ligand transfer requiring direct protein-membrane contact, that rates of ligand transfer are greater to more highly curved lipid vesicles, and that such rates are insensitive to the presence of anionic phospholipids in the acceptor membrane. These results point to hydrophobic features of alpha-TTP dominating the binding energy between protein and membrane.

Intelligible machine learning with malibu

malibu is an open-source machine learning work-bench developed in C/C++ for high-performance real-world applications, namely bioinformatics and medical informatics. It leverages third-party machine learning implementations for more robust bug-free software. This workbench handles several well-studied supervised machine learning problems including classification, regression, importance-weighted classification and multiple-instance learning. The malibu interface was designed to create reproducible experiments ideally run in a remote and/or command line environment. The software can be found at: http://proteomics.bioengr. uic.edu/malibu/index.html.

Lipid binding domains: more than simple lipid effectors

The spatial and temporal regulation of lipid molecules in cell membranes is a hallmark of cellular signaling and membrane trafficking events. Lipid-mediated targeting provides for strict control and versatility, because cell membranes harbor a large number of lipid molecules with variation in head group and acyl chain structures. Signaling and trafficking proteins contain a large number of modular domains that exhibit specific lipid binding properties and play a critical role in their localization and function. Nearly 20 years of research including structural, computational, biochemical and biophysical studies have demonstrated how these lipid-binding domains recognize their target lipid and achieve subcellular localization. The integration of this individual lipid-binding domain data in the context of the full-length proteins, macromolecular signaling complexes, and the lipidome is only beginning to be unraveled and represents a target of therapeutic development. This review brings together recent findings and classical concepts to concisely summarize the lipid-binding domain field while illustrating where the field is headed and how the gaps may be filled in with new technologies.

Disparate proteins use similar architectures to damage membranes

Membrane disruption can efficiently alter cellular function; indeed, pore-forming toxins (PFTs) are well known as important bacterial virulence factors. However, recent data have revealed that structures similar to those found in PFTs are found in membrane active proteins across disparate phyla. Many similarities can be identified only at the 3D-structural level. Of note, domains found in membrane-attack complex proteins of complement and perforin (MACPF) resemble cholesterol-dependent cytolysins from Gram-positive bacteria, and the Bcl family of apoptosis regulators share similar architectures with Escherichia coli pore-forming colicins. These and other correlations provide considerable help in understanding the structural requirements for membrane binding and pore formation.

Understanding structural features of microbial lipases--an overview

The structural elucidations of microbial lipases have been of prime interest since the 1980s. Knowledge of structural features plays an important role in designing and engineering lipases for specific purposes. Significant structural data have been presented for few microbial lipases, while, there is still a structure-deficit, that is, most lipase structures are yet to be resolved. A search for 'lipase structure' in the RCSB Protein Data Bank (http://www.rcsb.org/pdb/) returns only 93 hits (as of September 2007) and, the NCBI database (http://www.ncbi.nlm.nih.gov) reports 89 lipase structures as compared to 14719 core nucleotide records. It is therefore worthwhile to consider investigations on the structural analysis of microbial lipases. This review is intended to provide a collection of resources on the instrumental, chemical and bioinformatics approaches for structure analyses. X-ray crystallography is a versatile tool for the structural biochemists and is been exploited till today. The chemical methods of recent interests include molecular modeling and combinatorial designs. Bioinformatics has surged striking interests in protein structural analysis with the advent of innumerable tools. Furthermore, a literature platform of the structural elucidations so far investigated has been presented with detailed descriptions as applicable to microbial lipases. A case study of Candida rugosa lipase (CRL) has also been discussed which highlights important structural features also common to most lipases. A general profile of lipase has been vividly described with an overview of lipase research reviewed in the past.

Screening Outside the Catalytic Site: Inhibition of Macromolecular Inter-actions Through Structure-Based Virtual Ligand Screening Experiments

During these last 15 years, drug discovery strategies have essentially focused on identifying small molecules able to inhibit catalytic sites. However, other mechanisms could be targeted. Protein-protein interactions play crucial roles in a number of biological processes, and, as such, their disruption or stabilization is becoming an area of intense activity. Along the same line, inhibition of protein-membrane could be of major importance in several disease indications. Despite the many challenges associated with the development of such classes of interaction modulators, there has been considerable success in the recent years. Importantly, through the existence of protein hot-spots and the presence of druggable pockets at the macromolecular interfaces or in their vicinities, it has been possible to find small molecule effectors using a variety of screening techniques, including combined virtual ligand-in vitro screening strategy. Indeed such in silico-in vitro protocols emerge as the method of choice to facilitate our quest of novel drug-like compounds or of mechanistic probes aiming at facilitating the understanding of molecular reactions involved in the Health and Disease process. In this review, we comment recent successes of combined in silico-in vitro screening methods applied to modulating macromolecular interactions with a special emphasis on protein-membrane interactions.

MeTaDoR: a comprehensive resource for membrane targeting domains and their host proteins

MOTIVATION

Protein-lipid interactions play a central role in cellular signaling and membrane trafficking and at the core of these interactions are domains specialized in lipid binding and membrane targeting. Considering the importance of these domains, we have created MeTaDoR, a comprehensive resource dedicated to membrane targeting domains (MTDs).

RESULT

MeTaDoR begins with a brief introduction about all the important MTDs including their subcellular localization and structural features. Sequences of all known MTDs are then provided in two formats: standard Prosite format and a parsed tab-delimited format that provides a manually curated classification into binding or non-binding. Structures of all MTDs and host proteins known so far are provided with links to PDB and Pfam databases. Membrane-binding orientation of these proteins, whether experimentally determined or proposed, is also provided with links to the appropriate literature. To facilitate molecular dynamics studies of these proteins, the force-field parameters for many non-standard lipids that commonly interact with these proteins are also provided. Finally, an online server for predicting membrane-binding proteins and a search function with various search fields are included. The resource is publicly available and will be updated on a regular basis.

Design of protein membrane interaction inhibitors by virtual ligand screening, proof of concept with the C2 domain of factor V

Most orally bioavailable drugs on the market are competitive inhibitors of catalytic sites, but a significant number of targets remain undrugged, because their molecular functions are believed to be inaccessible to drug-like molecules. This observation specifically applies to the development of small-molecule inhibitors of macromolecular interactions such as protein-membrane interactions that have been essentially neglected thus far. Nonetheless, many proteins containing a membrane-targeting domain play a crucial role in health and disease, and the inhibition of such interactions therefore represents a very promising therapeutic strategy. In this study, we demonstrate the use of combined in silico structure-based virtual ligand screening and surface plasmon resonance experiments to identify compounds that specifically disrupt protein-membrane interactions. Computational analysis of several membrane-binding domains revealed they all contain a druggable pocket within their membrane-binding region. We applied our screening protocol to the second discoidin domain of coagulation factor V and screened >300,000 drug-like compounds in silico against two known crystal structure forms. For each C2 domain structure, the top 500 molecules predicted as likely factor V-membrane inhibitors were evaluated in vitro. Seven drug-like hits were identified, indicating that therapeutic targets that bind transiently to the membrane surface can be investigated cost-effectively, and that inhibitors of protein-membrane interactions can be designed.

The role of hydrophobic interactions in positioning of peripheral proteins in membranes

BACKGROUND

Three-dimensional (3D) structures of numerous peripheral membrane proteins have been determined. Biological activity, stability, and conformations of these proteins depend on their spatial positions with respect to the lipid bilayer. However, these positions are usually undetermined.

RESULTS

We report the first large-scale computational study of monotopic/peripheral proteins with known 3D structures. The optimal translational and rotational positions of 476 proteins are determined by minimizing energy of protein transfer from water to the lipid bilayer, which is approximated by a hydrocarbon slab with a decadiene-like polarity and interfacial regions characterized by water-permeation profiles. Predicted membrane-binding sites, protein tilt angles and membrane penetration depths are consistent with spin-labeling, chemical modification, fluorescence, NMR, mutagenesis, and other experimental studies of 53 peripheral proteins and peptides. Experimental membrane binding affinities of peripheral proteins were reproduced in cases that did not involve a helix-coil transition, specific binding of lipids, or a predominantly electrostatic association. Coordinates of all examined peripheral proteins and peptides with the calculated hydrophobic membrane boundaries, subcellular localization, topology, structural classification, and experimental references are available through the Orientations of Proteins in Membranes (OPM) database.

CONCLUSION

Positions of diverse peripheral proteins and peptides in the lipid bilayer can be accurately predicted using their 3D structures that represent a proper membrane-bound conformation and oligomeric state, and have membrane binding elements present. The success of the implicit solvation model suggests that hydrophobic interactions are usually sufficient to determine the spatial position of a protein in the membrane, even when electrostatic interactions or specific binding of lipids are substantial. Our results demonstrate that most peripheral proteins not only interact with the membrane surface, but penetrate through the interfacial region and reach the hydrocarbon interior, which is consistent with published experimental studies.

Mechanism of diacylglycerol-induced membrane targeting and activation of protein kinase Ctheta

Protein kinase C (PKC) is a novel PKC that plays a key role in T lymphocyte activation. PKC has been shown to be specifically recruited to the immunological synapse in response to T cell receptor activation. To understand the basis of its unique subcellular localization properties, we investigated the mechanism of in vitro and cellular sn-1,2-diacylglycerol (DAG)-mediated membrane binding of PKC. PKC showed phosphatidylserine selectivity in membrane binding and kinase action, which contributes to its translocation to the phosphatidylserine-rich plasma membrane in HEK293 cells. Unlike any other PKCs characterized so far, the isolated C1B domain of PKC had much higher affinity for DAG-containing membranes than the C1A domain. Also, the mutational analysis indicates that the C1B domain plays a predominant role in the DAG-induced membrane binding and activation of PKC. Furthermore, the Ca(2+)-independent C2 domain of PKC has significant affinity for anionic membranes, and the truncation of the C2 domain greatly enhanced the membrane affinity and enzyme activity of PKC. In addition, membrane binding properties of Y90E and Y90F mutants indicate that phosphorylation of Tyr(90) of the C2 domain enhances the affinity of PKC for model and cell membranes. Collectively, these results show that PKC has a unique membrane binding and activation mechanism that may account for its subcellular targeting properties.

High Cationic Charge and Bilayer Interface-Binding Helices in a Regulatory Lipid Glycosyltransferase,

In the prokaryote Acholeplasma laidlawii, membrane bilayer properties are sensed and regulated by two interface glycosyltransferases (GTs), synthesizing major nonbilayer- (alMGS GT) and bilayer-prone glucolipids. These enzymes are of similar structure, as many soluble GTs, but are sensitive to lipid charge and curvature stress properties. Multivariate and bioinformatic sequence analyses show that such interface enzymes, in relation to soluble ones of similar fold, are characterized by high cationic charge, certain distances between small and cationic amino acids, and by amphipathic helices. Varying surface contents of Lys/Arg pairs and Trp indicate different membrane-binding subclasses. A predicted potential (cationic) binding helix from alMGS was structurally verified by solution NMR and CD. The helix conformation was induced by a zwitterionic as well as anionic lipid environment, and the peptide was confined to the bilayer interface. Bilayer affinity of the peptide, analyzed by surface plasmon resonance, was higher than that for soluble membrane-seeking proteins/peptides and rose with anionic lipid content. Interface intercalation was supported by phase equilibria in membrane lipid mixtures, analyzed by 31P NMR and DSC. An analogous, potentially binding helix has a similar location in the structurally determined Escherichia coli cell wall precursor GT MurG. These two helices have little sequence conservation in alMGS and MurG homologues but maintain their amphipathic character. The evolutionary modification of the alMGS binding helix and its location close to the acceptor substrate site imply a functional importance in enzyme catalysis, potentially providing a mechanism by which glycolipid synthesis will be sensitive to membrane surface charge and intrinsic curvature strain.

Learning to translate sequence and structure to function: identifying DNA binding and membrane binding proteins

A protein's function depends in a large part on interactions with other molecules. With an increasing number of protein structures becoming available every year, a corresponding structural annotation approach identifying such interactions grows more expedient. At the same time, machine learning has gained popularity in bioinformatics providing robust annotation of genes and proteins without sequence homology. Here we have developed a general machine learning protocol to identify proteins that bind DNA and membrane. In general, there is no theory or even rule of thumb to pick the best machine learning algorithm. Thus, a systematic comparison of several classification algorithms known to perform well is investigated. Indeed, the boosted tree classifier is found to give the best performance, achieving 93% and 88% accuracy to discriminate non-homologous proteins that bind membrane and DNA, respectively, significantly outperforming all previously published works. We also attempted to address the importance of the attributes in function prediction and the relationships between relevant attributes. A graphical model based on boosted trees is applied to study the important features in discriminating DNA-binding proteins. In summary, the current protocol identified physical features important in DNA and membrane binding, rather than annotating function through sequence similarity.

Finding new structural and sequence attributes to predict possible disease association of single amino acid polymorphism (SAP)

MOTIVATION

The rapid accumulation of single amino acid polymorphisms (SAPs), also known as non-synonymous single nucleotide polymorphisms (nsSNPs), brings the opportunities and needs to understand and predict their disease association. Currently published attributes are limited, the detailed mechanisms governing the disease association of a SAP remain unclear and thus, further investigation of new attributes and improvement of the prediction are desired.

RESULTS

A SAP dataset was compiled from the Swiss-Prot variant pages. We extracted and demonstrated the effectiveness of several new biologically informative attributes including the structural neighbor profiles that describe the SAP's microenvironment, nearby functional sites that measure the structure-based and sequence-based distances between the SAP site and its nearby functional sites, aggregation properties that measure the likelihood of protein aggregation and disordered regions that consider whether the SAP is located in structurally disordered regions. The new attributes provided insights into the mechanisms of the disease association of SAPs. We built a support vector machines (SVMs) classifier employing a carefully selected set of new and previously published attributes. Through a strict protein-level 5-fold cross-validation, we attained an overall accuracy of 82.61%, and an MCC of 0.60. Moreover, a web server was developed to provide a user-friendly interface for biologists.

AVAILABILITY

The web server is available at http://sapred.cbi.pku.edu.cn/

Residue-level prediction of DNA-binding sites and its application on DNA-binding protein predictions

Protein-DNA interactions are crucial to many cellular activities such as expression-control and DNA-repair. These interactions between amino acids and nucleotides are highly specific and any aberrance at the binding site can render the interaction completely incompetent. In this study, we have three aims focusing on DNA-binding residues on the protein surface: to develop an automated approach for fast and reliable recognition of DNA-binding sites; to improve the prediction by distance-dependent refinement; use these predictions to identify DNA-binding proteins. We use a support vector machines (SVM)-based approach to harness the features of the DNA-binding residues to distinguish them from non-binding residues. Features used for distinction include the residue's identity, charge, solvent accessibility, average potential, the secondary structure it is embedded in, neighboring residues, and location in a cationic patch. These features collected from 50 proteins are used to train SVM. Testing is then performed on another set of 37 proteins, much larger than any testing set used in previous studies. The testing set has no more than 20% sequence identity not only among its pairs, but also with the proteins in the training set, thus removing any undesired redundancy due to homology. This set also has proteins with an unseen DNA-binding structural class not present in the training set. With the above features, an accuracy of 66% with balanced sensitivity and specificity is achieved without relying on homology or evolutionary information. We then develop a post-processing scheme to improve the prediction using the relative location of the predicted residues. Balanced success is then achieved with average sensitivity, specificity and accuracy pegged at 71.3%, 69.3% and 70.5%, respectively. Average net prediction is also around 70%. Finally, we show that the number of predicted DNA-binding residues can be used to differentiate DNA-binding proteins from non-DNA-binding proteins with an accuracy of 78%. Results presented here demonstrate that machine-learning can be applied to automated identification of DNA-binding residues and that the success rate can be ameliorated as more features are added. Such functional site prediction protocols can be useful in guiding consequent works such as site-directed mutagenesis and macromolecular docking.