Classifying G-protein coupled receptors with support vector machines

MOTIVATION: The enormous amount of protein sequence data uncovered by genome research has increased the demand for computer software that can automate the recognition of new proteins. We discuss the relative merits of various automated methods for recognizing G-Protein Coupled Receptors (GPCRs), a superfamily of cell membrane proteins. GPCRs are found in a wide range of organisms and are central to a cellular signalling network that regulates many basic physiological processes. They are the focus of a significant amount of current pharmaceutical research because they play a key role in many diseases. However, their tertiary structures remain largely unsolved. The methods described in this paper use only primary sequence information to make their predictions. We compare a simple nearest neighbor approach (BLAST), methods based on multiple alignments generated by a statistical profile Hidden Markov Model (HMM), and methods, including Support Vector Machines (SVMs), that transform protein sequences into fixed-length feature vectors. RESULTS: The last is the most computationally expensive method, but our experiments show that, for those interested in annotation-quality classification, the results are worth the effort. In two-fold cross-validation experiments testing recognition of GPCR subfamilies that bind a specific ligand (such as a histamine molecule), the errors per sequence at the Minimum Error Point (MEP) were 13.7% for multi-class SVMs, 17.1% for our SVMtree method of hierarchical multi-class SVM classification, 25.5% for BLAST, 30% for profile HMMs, and 49% for classification based on nearest neighbor feature vector Kernel Nearest Neighbor (kernNN). The percentage of true positives recognized before the first false positive was 65% for both SVM methods, 13% for BLAST, 5% for profile HMMs and 4% for kernNN.     

Translocation mechanism of long sugar chains across the maltoporin membrane channel

Maltoporin allows permeation of long maltodextrin chains. It tightly binds the amphiphilic sugar, offering both hydrophobic interactions with a helical lane of aromatic residues and H bonds with ionic side chains. The minimum-energy path of maltohexaose translocation is obtained by the conjugate peak refinement method, which optimizes a continuous string of conformers without applying constraints. This reveals that the protein is passive while the sugar glides screw-like along the aromatic lane. Near instant switching of sugar hydroxyl H bond partners results in two small energy barriers (of approximately 4 kcal/mol each) during register shift by one glucosyl unit, in agreement with a kinetic analysis of experimental dissociation rates for varying sugar chain lengths. Thus, maltoporin functions like an efficient translocation "enzyme," and the slow rate of the register shift (approximately 1/ms) is due to high collisional friction.     

Validity of Gō models: comparison with a solvent-shielded empirical energy decomposition

Do Gō-type model potentials provide a valid approach for studying protein folding? They have been widely used for this purpose because of their simplicity and the speed of simulations based on their use. The essential assumption in such models is that only contact interactions existing in the native state determine the energy surface of a polypeptide chain, even for non-native configurations sampled along folding trajectories. Here we use an all-atom molecular mechanics energy function to investigate the adequacy of Gō-type potentials. We show that, although the contact approximation is accurate, non-native contributions to the energy can be significant. The assumed relation between residue-residue interaction energies and the number of contacts between them is found to be only approximate. By contrast, individual residue energies correlate very well with the number of contacts. The results demonstrate that models based on the latter should give meaningful results (e.g., as used to interpret phi values), whereas those that depend on the former are only qualitative, at best.     

Competition between C-terminal tyrosine and nicotinamide modulates pyridine nucleotide affinity and specificity in plant ferredoxin-NADP(+) reductase

Chloroplast ferredoxin-NADP(+) reductase has a 32,000-fold preference for NADPH over NADH, consistent with its main physiological role of NADP(+) photoreduction for de novo carbohydrate biosynthesis. Although it is distant from the 2'-phosphoryl group of NADP(+), replacement of the C-terminal tyrosine (Tyr(308) in the pea enzyme) by Trp, Phe, Gly, and Ser produced enzyme forms in which the preference for NADPH over NADH was decreased about 2-, 10-, 300-, and 400-fold, respectively. Remarkably, in the case of the Y308S mutant, the k(cat) value for the NADH-dependent activity approached that of the NADPH-dependent activity of the wild-type enzyme. Furthermore, difference spectra of the NAD(+) complexes revealed that the nicotinamide ring of NAD(+) binds at nearly full occupancy in the active site of both the Y308G and Y308S mutants. These results correlate well with the k(cat) values obtained with these mutants in the NADH-ferricyanide reaction. The data presented support the hypothesis that specific recognition of the 2'-phosphate group of NADP(H) is required but not sufficient to ensure a high degree of discrimination against NAD(H) in ferredoxin-NADP(+) reductase. Thus, the C-terminal tyrosine enhances the specificity of the reductase for NADP(H) by destabilizing the interaction of a moiety common to both coenzymes, i.e. the nicotinamide.     

Wave propagation in cardiac tissue and effects of intracellular calcium dynamics (computer simulation study)

Computer simulation using Luo-Rudy I1 model of ventricular myocyte showed that intracellular calcium dynamics become irregular in case of high rate stimulation. This causes the transition from stationary to nonstationary spiral wave and its breakup in 2D model of cardiac tissue. Obtained results suggest how ventricular fibrillation may occur due to the abnormalities of intracellular calcium dynamics. The short review of existing cardiac cell models with calcium dynamics is presented.     

Self-masking in an intact ERM-merlin protein: an active role for the central alpha-helical domain

Ezrin/radixin/moesin (ERM) family members provide a regulated link between the cortical actin cytoskeleton and the plasma membrane to govern membrane structure and organization. Here, we report the crystal structure of intact insect moesin, revealing that its essential yet previously uncharacterized alpha-helical domain forms extensive interactions with conserved surfaces of the band four-point-one/ezrin/radixin/moesin (FERM) domain. These interdomain contacts provide a functional explanation for how PIP(2) binding and tyrosine phosphorylation of ezrin lead to activation, and provide an understanding of previously enigmatic loss-of-function missense mutations in the tumor suppressor merlin. Sequence conservation and biochemical results indicate that this structure represents a complete model for the closed state of all ERM-merlin proteins, wherein the central alpha-helical domain is an active participant in an extensive set of inhibitory interactions that can be unmasked, in a rheostat-like manner, by coincident regulatory factors that help determine cell polarity and membrane structure.