Recent improvements in prediction of protein structure by global optimization of a potential energy function

Recent improvements of a hierarchical ab initio or de novo approach for predicting both α and β structures of proteins are described. The united-residue energy function used in this procedure includes multibody interactions from a cumulant expansion of the free energy of polypeptide chains, with their relative weights determined by Z-score optimization. The critical initial stage of the hierarchical procedure involves a search of conformational space by the conformational space annealing (CSA) method, followed by optimization of an all-atom model. The procedure was assessed in a recent blind test of protein structure prediction (CASP4). The resulting lowest-energy structures of the target proteins (ranging in size from 70 to 244 residues) agreed with the experimental structures in many respects. The entire experimental structure of a cyclic α-helical protein of 70 residues was predicted to within 4.3 Å α-carbon (Cα) rms deviation (rmsd) whereas, for other α-helical proteins, fragments of roughly 60 residues were predicted to within 6.0 Å Cα rmsd. Whereas β structures can now be predicted with the new procedure, the success rate for α/β- and β-proteins is lower than that for α-proteins at present. For the β portions of α/β structures, the Cα rmsd's are less than 6.0 Å for contiguous fragments of 30–40 residues; for one target, three fragments (of length 10, 23, and 28 residues, respectively) formed a compact part of the tertiary structure with a Cα rmsd less than 6.0 Å. Overall, these results constitute an important step toward the ab initio prediction of protein structure solely from the amino acid sequence.

Global properties of the mapping between local amino acid sequence and local structure in proteins.

Local protein structure prediction efforts have consistently failed to exceed approximately 70% accuracy. We characterize the degeneracy of the mapping from local sequence to local structure responsible for this failure by investigating the extent to which similar sequence segments found in different proteins adopt similar three-dimensional structures. Sequence segments 3-15 residues in length from 154 different protein families are partitioned into neighborhoods containing segments with similar sequences using cluster analysis. The consistency of the sequence-to-structure mapping is assessed by comparing the local structures adopted by sequence segments in the same neighborhood in proteins of known structure. In the 154 families, 45% and 28% of the positions occur in neighborhoods in which one and two local structures predominate, respectively. The sequence patterns that characterize the neighborhoods in the first class probably include virtually all of the short sequence motifs in proteins that consistently occur in a particular local structure. These patterns, many of which occur in transitions between secondary structural elements, are an interesting combination of previously studied and novel motifs. The identification of sequence patterns that consistently occur in one or a small number of local structures in proteins should contribute to the prediction of protein structure from sequence.

Prediction of protein folding class using global description of amino acid sequence.

We present a method for predicting protein folding class based on global protein chain description and a voting process. Selection of the best descriptors was achieved by a computer-simulated neural network trained on a data base consisting of 83 folding classes. Protein-chain descriptors include overall composition, transition, and distribution of amino acid attributes, such as relative hydrophobicity, predicted secondary structure, and predicted solvent exposure. Cross-validation testing was performed on 15 of the largest classes. The test shows that proteins were assigned to the correct class (correct positive prediction) with an average accuracy of 71.7%, whereas the inverse prediction of proteins as not belonging to a particular class (correct negative prediction) was 90-95% accurate. When tested on 254 structures used in this study, the top two predictions contained the correct class in 91% of the cases.

Protein contact prediction using patterns of correlation

We describe a new method for using neural networks to predict residue contact pairs in a protein. The main inputs to the neural network are a set of 25 measures of correlated mutation between all pairs of residues in two windows of size 5 centered on the residues of interest. While the individual pair-wise correlations are a relatively weak predictor of contact, by training the network on windows of correlation the accuracy of prediction is significantly improved. The neural network is trained on a set of 100 proteins and then tested on a disjoint set of 1033 proteins of known structure. An average predictive accuracy of 21.7% is obtained taking the best L/2 predictions for each protein, where L is the sequence length. Taking the best L/10 predictions gives an average accuracy of 30.7%. The predictor is also tested on a set of 59 proteins from the CASP5 experiment. The accuracy is found to be relatively consistent across different sequence lengths, but to vary widely according to the secondary structure. Predictive accuracy is also found to improve by using multiple sequence alignments containing many sequences to calculate the correlations. (C) 2004 Wiley-Liss, Inc.

Better prediction of protein contact number using a support vector regression analysis of amino acid sequence

Background: Protein tertiary structure can be partly characterized via each amino acid's contact number measuring how residues are spatially arranged. The contact number of a residue in a folded protein is a measure of its exposure to the local environment, and is defined as the number of C-beta atoms in other residues within a sphere around the C-beta atom of the residue of interest. Contact number is partly conserved between protein folds and thus is useful for protein fold and structure prediction. In turn, each residue's contact number can be partially predicted from primary amino acid sequence, assisting tertiary fold analysis from sequence data. In this study, we provide a more accurate contact number prediction method from protein primary sequence. Results: We predict contact number from protein sequence using a novel support vector regression algorithm. Using protein local sequences with multiple sequence alignments (PSI-BLAST profiles), we demonstrate a correlation coefficient between predicted and observed contact numbers of 0.70, which outperforms previously achieved accuracies. Including additional information about sequence weight and amino acid composition further improves prediction accuracies significantly with the correlation coefficient reaching 0.73. If residues are classified as being either contacted or non-contacted, the prediction accuracies are all greater than 77%, regardless of the choice of classification thresholds. Conclusion: The successful application of support vector regression to the prediction of protein contact number reported here, together with previous applications of this approach to the prediction of protein accessible surface area and B-factor profile, suggests that a support vector regression approach may be very useful for determining the structure-function relation between primary sequence and higher order consecutive protein structural and functional properties.

Prediction of subcellular localization using sequence-biased recurrent networks

Motivation: Targeting peptides direct nascent proteins to their specific subcellular compartment. Knowledge of targeting signals enables informed drug design and reliable annotation of gene products. However, due to the low similarity of such sequences and the dynamical nature of the sorting process, the computational prediction of subcellular localization of proteins is challenging. Results: We contrast the use of feed forward models as employed by the popular TargetP/SignalP predictors with a sequence-biased recurrent network model. The models are evaluated in terms of performance at the residue level and at the sequence level, and demonstrate that recurrent networks improve the overall prediction performance. Compared to the original results reported for TargetP, an ensemble of the tested models increases the accuracy by 6 and 5% on non-plant and plant data, respectively.

Prediction of Phosphorylation Sites from Protein Sequences

Phosphorylation is amongst the most crucial and well-studied post-translational modifications. It is involved in multiple cellular processes which makes phosphorylation prediction vital for understanding protein functions. However, wet-lab techniques are labour and time intensive. Thus, computational tools are required for efficiency. This project aims to provide a novel way to predict phosphorylation sites from protein sequences by adding flexibility and Sezerman Grouping amino acid similarity measure to previous methods, as discovering new protein sequences happens at a greater rate than determining protein structures. The predictor – NOPAY - relies on Support Vector Machines (SVMs) for classification. The features include amino acid encoding, amino acid grouping, predicted secondary structure, predicted protein disorder, predicted protein flexibility, solvent accessibility, hydrophobicity and volume. As a result, we have managed to improve phosphorylation prediction accuracy for Homo sapiens by 3% and 6.1% for Mus musculus. Sensitivity at 99% specificity was also increased by 6% for Homo sapiens and for Mus musculus by 5% on independent test sets. In this study, we have managed to increase phosphorylation prediction accuracy for Homo sapiens and Mus musculus. When there is enough data, future versions of the software may also be able to predict other organisms.

The Putative Leishmania Telomerase Rna (leishter) Undergoes Trans-splicing And Contains A Conserved Template Sequence.

Telomerase RNAs (TERs) are highly divergent between species, varying in size and sequence composition. Here, we identify a candidate for the telomerase RNA component of Leishmania genus, which includes species that cause leishmaniasis, a neglected tropical disease. Merging a thorough computational screening combined with RNA-seq evidence, we mapped a non-coding RNA gene localized in a syntenic locus on chromosome 25 of five Leishmania species that shares partial synteny with both Trypanosoma brucei TER locus and a putative TER candidate-containing locus of Crithidia fasciculata. Using target-driven molecular biology approaches, we detected a ∼2,100 nt transcript (LeishTER) that contains a 5' spliced leader (SL) cap, a putative 3' polyA tail and a predicted C/D box snoRNA domain. LeishTER is expressed at similar levels in the logarithmic and stationary growth phases of promastigote forms. A 5'SL capped LeishTER co-immunoprecipitated and co-localized with the telomerase protein component (TERT) in a cell cycle-dependent manner. Prediction of its secondary structure strongly suggests the existence of a bona fide single-stranded template sequence and a conserved C[U/C]GUCA motif-containing helix II, representing the template boundary element. This study paves the way for further investigations on the biogenesis of parasite TERT ribonucleoproteins (RNPs) and its role in parasite telomere biology.

Solution structure of amyloid beta-peptide(1-40) in a water-micelle environment. Is the membrane-spanning domain where we think it is?

The three-dimensional solution structure of the 40 residue amyloid beta-peptide, A beta(1-40), has been determined using NMR spectroscopy at pH 5.1, in aqueous sodium dodecyl sulfate (SDS) micelles, In this environment, which simulates to some extent a water-membrane medium, the peptide is unstructured between residues 1 and 14 which are mainly polar and likely solvated by water. However, the rest of the protein adopts an alpha-helical conformation between residues 15 and 36 with a kink or hinge at 25-27. This largely hydrophobic region is likely solvated by SDS. Based on the derived structures, evidence is provided in support of a possible new location for the transmembrane domain of A beta within the amyloid precursor protein (APP). Studies between pH 4.2 and 7.9 reveal a pH-dependent helix-coil conformational switch. At the lower pH values, where the carboxylate residues are protonated, the helix is uncharged, intact, and lipid-soluble. As the pH increases above 6.0, part of the helical region (15-24) becomes less structured, particularly near residues E22 and D23 where deprotonation appears to facilitate unwinding of the helix. This pH-dependent unfolding to a random coil conformation precedes any tendency of this peptide to aggregate to a beta-sheet as the pH increases. The structural biology described herein for A beta(1-40) suggests that (i) the C-terminal two-thirds of the peptide is an alpha-helix in membrane-like environments, (ii) deprotonation of two acidic amino acids in the helix promotes a helix-coil conformational transition that precedes aggregation, (iii) a mobile hinge exists in the helical region of A beta(1-40) and this may be relevant to its membrane-inserting properties and conformational rearrangements, and (iv) the location of the transmembrane domain of amyloid precursor proteins may be different from that accepted in the Literature. These results may provide new insight to the structural properties of amyloid beta-peptides of relevance to Alzheimer's disease.

Solution structure of methionine-oxidized amyloid beta-peptide (1-40). Does oxidation affect conformational switching?

The solution structure of A beta(1-40)Met(O), the methionine-oxidized form of amyloid beta-peptide A beta(1-40), has been investigated by CD and NMR spectroscopy. Oxidation of Met35 may have implications in the aetiology of Alzheimer's disease. Circular dichroism experiments showed that whereas A beta(1-40) and A beta(1-40)Met(O) both adopt essentially random coil structures in water (pH 4) at micromolar concentrations, the former aggregates within several days while the latter is stable for at least 7 days under these conditions. This remarkable difference led us to determine the solution structure of A beta(1-40)Met(O) using H-1 NMR spectroscopy. In a water-SDS micelle medium needed to solubilize both peptides at the millimolar concentrations required to measure NMR spectra, chemical shift and NOE data for A beta(1-40)Met(O) strongly suggest the presence of a helical region between residues 16 and 24. This is supported by slow H-D exchange of amide protons in this region and by structure calculations using simulated annealing with the program XPLOR. The remainder of the structure is relatively disordered. Our previously reported NMR data for A beta(1-40) in the same solvent shows that helices are present over residues 15-24 (helix 1) and 28-36 (helix 2), Oxidation of Met35 thus causes a local and selective disruption of helix 2. In addition to this helix-coil rearrangement in aqueous micelles, the CD data show that oxidation inhibits a coil-to-beta-sheet transition in water. These significant structural rearrangements in the C-terminal region of A beta may be important clues to the chemistry and biology of A beta(1-40) and A beta(1-42).

Three-dimensional solution structure of alpha-conotoxin MII by NMR spectroscopy: Effects of solution environment on helicity

alpha-Conotoxin MII, a 16-residue polypeptide from the venom of the piscivorous cone snail Conus magus, is a potent and highly specific blocker of mammalian neuronal nicotinic acetylcholine receptors composed of alpha 3 beta 2 subunits. The role of this receptor type in the modulation of neurotransmitter release and its relevance to the problems of addiction and psychosis emphasize the importance of a structural understanding of the mode of interaction of MII with the alpha 3 beta 2 interface. Here we describe the three-dimensional solution structure of MIT determined using 2D H-1 NMR spectroscopy. Structural restraints consisting of 376 interproton distances inferred from NOEs and 12 dihedral restraints derived from spin-spin coupling constants were used as input for simulated annealing calculations and energy minimization in the program X-PLOR. The final set of 20 structures is exceptionally well-defined with mean pairwise rms differences over the whole molecule of 0.07 Angstrom for the backbone atoms and 0.34 Angstrom for all heavy atoms. MII adopts a compact structure incorporating a central segment of alpha-helix and beta-turns at the N- and C-termini. The molecule is stabilized by two disulfide bonds, which provide cross-links between the N-terminus and both the middle and C-terminus of the structure. The susceptibility of the structure to conformational change was examined using several different solvent conditions. While the global fold of MII remains the same, the structure is stabilized in a more hydrophobic environment provided by the addition of acetonitrile or trifluoroethanol to the aqueous solution. The distribution of amino acid side chains in MII creates distinct hydrophobic and polar patches on its surface that may be important for the specific interaction with the alpha 3 beta 2 neuronal nAChR. A comparison of the structure of MII with other neuronal-specific alpha-conotoxins provides insights into their mode of interaction with these receptors.

Structure-activity studies of conantokins as human N-methyl-D-aspartate receptor modulators

The activities of conantokin-G (con-G), conantokin-T (con-T), and several novel analogues have been studied using polyamine enhancement of [H-3]MK-801 binding to human glutamate-N-methyl-D-aspartate (NMDA) receptors, and their structures have been examined using CD and H-1 NMR spectroscopy. The potencies of con-G[A7], con-G, and con-T as noncompetitive inhibitors of spermine-enhanced [H-3]MK-801 binding to NMDA receptor obtained from human brain tissue are similar to those obtained using rat brain tissue. The secondary structure and activity of con-G are found to be highly sensitive to amino acid substitution and modification. NMR chemical shift data indicate that con-G, con-G[D8,D17], and con-G[A7] have similar conformations in the presence of Ca2+. This consists of a helix for residues 2-16, which is kinked in the vicinity of Gla10. This is confirmed by 3D structure calculations on con-G[A7]. Restraining this helix in a linear form (i.e., con-G[A7,E10-K13]) results in a minor reduction in potency. Incorporation of a 7-10 salt-bridge replacement (con-G[K7-E10]) prevents helix formation in aqueous solution and produces a peptide with low potency. Peptides with the Leu5-Tyr5 substitution also have low potencies (con-G[Y5,A7] and con-G[Y5,K7]) indicating that Leu5 in con-G is important for full antagonist behavior. We have also shown that the Gla-Ala7 substitution increases potency, whereas the Gla-Lys7 substitution has no effect. Con-G and con-G[K7] both exhibit selectivity between NMDA subtypes from mid-frontal and superior temporal gyri, but not between sensorimotor and mid-frontal gyri. Asn8 and/or Asn17 appear to be important for the ability of con-G to function as an inhibitor of polyamine-stimulated [3H]MK-801 binding, but not in maintaining secondary structure. The presence of Ca2+ does not increase the potencies of con-G and con-T for NMDA receptors but does stabilize the helical structures of con-G, con-G[D8,D17], and, to a lesser extent, con-G[A7]. The NMR data support the existence of at least two independent Ca2+-chelating sites in con-G, one involving Gla7 and possibly Gla3 and the other likely to involve Gla10 and/or Gla14.

Structure-activity relationships of omega-conotoxins MVIIA, MVIIC and 14 loop splice hybrids at N and P/Q-type calcium channels

The omega-conotoxins are a set of structurally related, four-loop, six cysteine containing peptides, that have a range of selectivities for different subtypes of the voltage-sensitive calcium channel (VSCC). To investigate the basis of the selectivity displayed by these peptides, we have studied the binding affinities of two naturally occurring omega-conotoxins, MVIIA and MVIIC and a series of 14 MVIIA/MVIIC loop hybrids using radioligand binding assays for N and P/Q-type Ca2+ channels in rat brain tissue. A selectivity profile was developed from the ratio of relative potencies at N-type VSCCs (using [I-125]GVIA radioligand binding assays) and P/Q-type VSCCs (using [I-125]MVIIC radioligand binding assays). in these peptides, loops 2 and 4 make the greatest contribution to VSCC subtype selectivity, while the effects of loops 1 and 3 are negligible. Peptides with homogenous combinations of loop 2 and 4 display clear selectivity preferences, while those with heterogeneous combinations of loops 2 and 4 are less discriminatory. H-1 NMR spectroscopy revealed that the global folds of MVIIA, MVIIC and the 14 loop hybrid peptides were similar; however, several differences in local structure were identified. Based on the binding data and the 3D structures of MVIIA, GVIA and MVIIC, we have developed a preliminary pharmacophore based on the omega-conotoxin residues most Likely to interact with the N-type VSCC. (C) 1999 Academic Press.