Morten Källberg

Template-based protein structure modeling using the RaptorX web server

A key challenge of modern biology is to uncover the functional role of the protein entities that compose cellular proteomes. To this end, the availability of reliable three-dimensional atomic models of proteins is often crucial. This protocol presents a community-wide web-based method using RaptorX (http://raptorx.uchicago.edu/) for protein secondary structure prediction, template-based tertiary structure modeling, alignment quality assessment and sophisticated probabilistic alignment sampling. RaptorX distinguishes itself from other servers by the quality of the alignment between a target sequence and one or multiple distantly related template proteins (especially those with sparse sequence profiles) and by a novel nonlinear scoring function and a probabilistic-consistency algorithm. Consequently, RaptorX delivers high-quality structural models for many targets with only remote templates. At present, it takes RaptorX ∼35 min to finish processing a sequence of 200 amino acids. Since its official release in August 2011, RaptorX has processed ∼6,000 sequences submitted by ∼1,600 users from around the world.

RaptorX server: A Resource for Template-Based Protein Structure Modeling

Assigning functional properties to a newly discovered protein is a key challenge in modern biology. To this end, computational modeling of the three-dimensional atomic arrangement of the amino acid chain is often crucial in determining the role of the protein in biological processes. We present a community-wide web-based protocol, RaptorX server ( http://raptorx.uchicago.edu ), for automated protein secondary structure prediction, template-based tertiary structure modeling, and probabilistic alignment sampling.Given a target sequence, RaptorX server is able to detect even remotely related template sequences by means of a novel nonlinear context-specific alignment potential and probabilistic consistency algorithm. Using the protocol presented here it is thus possible to obtain high-quality structural models for many target protein sequences when only distantly related protein domains have experimentally solved structures. At present, RaptorX server can perform secondary and tertiary structure prediction of a 200 amino acid target sequence in approximately 30 min.

Cholesterol modulates cell signaling and protein networking by specifically interacting with PDZ domain-containing scaffold proteins

Cholesterol is known to modulate the physical properties of cell membranes but its direct involvement in cellular signaling has not been thoroughly investigated. Here we show that cholesterol specifically binds many PDZ domains found in scaffold proteins, including the N-terminal PDZ domain of NHERF1/EBP50. This modular domain has a cholesterol-binding site topologically distinct from its canonical protein-binding site and serves as a dual specificity domain that bridges the membrane and juxta-membrane signaling complexes. Disruption of the cholesterol binding activity of NHERF1 largely abrogates its dynamic colocalization with and activation of cystic fibrosis transmembrane conductance regulator, one of its binding partners in the plasma membrane of mammalian cells. At least seven more PDZ domains from other scaffold proteins also bind cholesterol and have cholesterol-binding sites, suggesting that cholesterol modulates cell signaling through direct interactions with these scaffold proteins. This mechanism may provide an alternative explanation for the formation of signaling platforms in cholesterol-rich membrane domains.

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.

Molecular Basis of the Potent Membrane-remodeling Activity of the Epsin 1 N-terminal Homology Domain

The mechanisms by which cytosolic proteins reversibly bind the membrane and induce the curvature for membrane trafficking and remodeling remain elusive. The epsin N-terminal homology (ENTH) domain has potent vesicle tubulation activity despite a lack of intrinsic molecular curvature. EPR revealed that the N-terminal α-helix penetrates the phosphatidylinositol 4,5-bisphosphate-containing membrane at a unique oblique angle and concomitantly interacts closely with helices from neighboring molecules in an antiparallel orientation. The quantitative fluorescence microscopy showed that the formation of highly ordered ENTH domain complexes beyond a critical size is essential for its vesicle tubulation activity. The mutations that interfere with the formation of large ENTH domain complexes abrogated the vesicle tubulation activity. Furthermore, the same mutations in the intact epsin 1 abolished its endocytic activity in mammalian cells. Collectively, these results show that the ENTH domain facilitates the cellular membrane budding and fission by a novel mechanism that is distinct from that proposed for BAR domains.

Mechanical Signaling on the Single Protein Level Studied Using Steered Molecular Dynamics

Efficient communication between the cell and its external environment is of the utmost importance to the function of multicellular organisms. While signaling events can be generally characterized as information exchange by means of controlled energy conversion, research efforts have hitherto mainly been concerned with mechanisms involving chemical and electrical energy transfer. Here, we review recent computational efforts addressing the function of mechanical force in signal transduction. Specifically, we focus on the role of steered molecular dynamics (SMD) simulations in providing details at the atomic level on a group of protein domains, which play a fundamental role in signal exchange by responding properly to mechanical strain. We start by giving a brief introduction to the SMD technique and general properties of mechanically stable protein folds, followed by specific examples illustrating three general regimes of signal transfer utilizing mechanical energy: purely mechanical, mechanical to chemical, and chemical to mechanical. Whenever possible the physiological importance of the example at hand is stressed to highlight the diversity of the processes in which mechanical signaling plays a key role. We also provide an overview of future challenges and perspectives for this rapidly developing field.

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

Structural feature extraction protocol for classifying reversible membrane binding protein domains

Machine learning based classification protocols for automated function annotation of protein structures have in many instances proven superior to simpler sequence based procedures. Here we present an automated method for extracting features from protein structures by construction of surface patches to be used in such protocols. The utility of the developed patch-growing procedure is exemplified by its ability to identify reversible membrane binding domains from the C1, C2, and PH families.

Understanding Protein-DNA Interactions through Dynamics

Transcriptional regulation is a key factor in controlling proper cellular behavior. For this reason, so-called regulation networks (quantifying the molecular interactions controlling the transcription), have been heavily studied. One goal is to enrich these networks through in silico identification of DNA-binding proteins and their respective binding sites. Often such work assumes a specific distance between atoms as constituting an interaction and construct models based on this assumption. However, this ad hoc rule fails to account for many of the complexities that lie behind the physical nature of binding. We present a framework for studying these interactions in more realistic settings accounting for both overall energy and dynamics of protein-DNA complex. We demonstrate that short molecular dynamics simulations better characterize biomolecular interactions and that a better definition of interactions improves the prediction of protein-DNA docking. Specifically, interacting residues are identified through the analysis of MD energy functions and results are compared with published experimental results. Further, we show how our novel definition of DNA-binding can be used for constructing improved machine learning classifiers for automatic identification of DNA-binding residues.

A Machine Learning Protocol for Distinguish Intra-domain Peripheral Membrane Targeting Properties using Sequence and Structure

Peripheral membrane-targeting proteins can associate with membranes in a reversible manner, allowing for the transient localization of these proteins to specific intracellular sites. Due to this property, such proteins are often found to be important players in both signal transduction and protein trafficking processes. A number of domain families, for example C1-, C2-, and PH-domains, have been found to be of great importance in driving this type of association, however, no specific sequence motifs eluding to the targeting properties have been identified in these families. For this reason, a simple procedure based on sequence similarity alone will not be effective in computational function annotation.We present a machine learning protocol for distinguishing intra-family membrane-targeting properties. The protocol is based on features obtained from both sequence and structure allowing for the incorporation of both statistics obtained from the entire domain family as well as physical quantities specific to each domain. First, values for residue conservation in targeting versus non-targeting domains are calculated. Second, properties such electrostatics and surface hydrophobicity of the domain are quantified by defining patches of similar values on the solvent exposed surface of the structure. Based on these features we construct a classification model for each family and furthermore compare the performance of a number of algorithms in this problem domain.Finally, we explore the interdependence of the features in determining the membrane-targeting properties of each family through the use of alternating decision trees, drawing out the specific targeting properties for each family.