Robert W. Williams

PONDR-FIT: A Meta-Predictor of Intrinsically Disordered Amino Acids

Protein intrinsic disorder is becoming increasingly recognized in proteomics research. While lacking structure, many regions of disorder have been associated with biological function. There are many different experimental methods for characterizing intrinsically disordered proteins and regions; nevertheless, the prediction of intrinsic disorder from amino acid sequence remains a useful strategy especially for many large-scale proteomic investigations. Here we introduced a consensus artificial neural network (ANN) prediction method, which was developed by combining the outputs of several individual disorder predictors. By eight-fold cross-validation, this meta-predictor, called PONDR-FIT, was found to improve the prediction accuracy over a range of 3 to 20% with an average of 11% compared to the single predictors, depending on the datasets being used. Analysis of the errors shows that the worst accuracy still occurs for short disordered regions with less than ten residues, as well as for the residues close to order/disorder boundaries. Increased understanding of the underlying mechanism by which such meta-predictors give improved predictions will likely promote the further development of protein disorder predictors. Access to PONDR-FIT is available at www.disprot.org.

Secondary structure predictions and medium range interactions

Several authors have proposed that predictions of protein secondary structure derived from statistical information about the known structures can be improved when information about neighboring residues participating in short and medium range interactions is included. A substantial improvement shown here indicates that current methods of including this information are not more successful than methods that do not. Evaluations of the Chou and Fasman method (Adv. Enzymol. 47 (1978) 45-148), that does not include information about interactions (except in averaging), have shown it to be about 49% correct for three states (helix, beta-sheet and undefined). In comparison, the method of Garnier et al. (J. Mol. Biol. 120 (1978) 97-120), that explicitly includes information about neighboring residues, has an accuracy of 57% residues correct for three states. However, we have obtained an 8% improvement for predictions of secondary structure based on the algorithm by Chou and Fasman. The improvements are obtained by eliminating many rules and by choosing the best decision constants for structure assignments. The simplified method described here is 57% correct for three states using preference values calculated in 1978.

Mutations in the yeast LCB1 and LCB2 genes, including those corresponding to the hereditary sensory neuropathy type I mutations, dominantly inactivate serine palmitoyltransferase

It was recently demonstrated that mutations in the human SPTLC1 gene, encoding the Lcb1p subunit of serine palmitoyltransferase (SPT), cause hereditary sensory neuropathy type I (1, 2). As a member of the subfamily of pyridoxal 5′-phosphate enzymes known as the α-oxoamine synthases, serine palmitoyltransferase catalyzes the committed step of sphingolipid synthesis. The residues that are mutated to cause hereditary sensory neuropathy type I reside in a highly conserved region of Lcb1p that is predicted to be a catalytic domain of Lcb1p on the basis of alignments with other members of the α-oxoamine synthase family. We found that the corresponding mutations in the LCB1 gene ofSaccharomyces cerevisiae reduce serine palmitoyltransferase activity. These mutations are dominant and decrease serine palmitoyltransferase activity by 50% when the wild-type and mutantLCB1 alleles are coexpressed. We also show that serine palmitoyltransferase is an Lcb1p·Lcb2p heterodimer and that the mutated Lcb1p proteins retain their ability to interact with Lcb2p. Modeling studies suggest that serine palmitoyltransferase is likely to have a single active site that lies at the Lcb1p·Lcb2p interface and that the mutations in Lcb1p reside near the lysine in Lcb2p that is expected to form the Schiffs base with the pyridoxal 5′-phosphate cofactor. Furthermore, mutations in this lysine and in a histidine residue that is also predicted to be important for pyridoxal 5′-phosphate binding to Lcb2p also dominantly inactivate SPT similar to the hereditary sensory neuropathy type 1-like mutations in Lcb1p.

Archaic chaos: intrinsically disordered proteins in Archaea

BackgroundMany proteins or their regions known as intrinsically disordered proteins (IDPs) and intrinsically disordered regions (IDRs) lack unique 3D structure in their native states under physiological conditions yet fulfill key biological functions. Earlier bioinformatics studies showed that IDPs and IDRs are highly abundant in different proteomes and carry out mostly regulatory functions related to molecular recognition and signal transduction. Archaea belong to an intriguing domain of life whose members, being microbes, are characterized by a unique mosaic-like combination of bacterial and eukaryotic properties and include inhabitants of some of the most extreme environments on the planet. With the expansion of the archaea genome data (more than fifty archaea species from five different phyla are known now), and with recent improvements in the accuracy of intrinsic disorder prediction, it is time to re-examine the abundance of IDPs and IDRs in the archaea domain.ResultsThe abundance of IDPs and IDRs in 53 archaea species is analyzed. The amino acid composition profiles of these species are generally quite different from each other. The disordered content is highly species-dependent. Thermoproteales proteomes have 14% of disordered residues, while in Halobacteria, this value increases to 34%. In proteomes of these two phyla, proteins containing long disordered regions account for 12% and 46%, whereas 4% and 26% their proteins are wholly disordered. These three measures of disorder content are linearly correlated with each other at the genome level. There is a weak correlation between the environmental factors (such as salinity, pH and temperature of the habitats) and the abundance of intrinsic disorder in Archaea, with various environmental factors possessing different disorder-promoting strengths. Harsh environmental conditions, especially those combining several hostile factors, clearly favor increased disorder content. Intrinsic disorder is highly abundant in functional Pfam domains of the archaea origin. The analysis based on the disordered content and phylogenetic tree indicated diverse evolution of intrinsic disorder among various classes and species of Archaea.ConclusionsArchaea proteins are rich in intrinsic disorder. Some of these IDPs and IDRs likely evolve to help archaea to accommodate to their hostile habitats. Other archaean IDPs and IDRs possess crucial biological functions similar to those of the bacterial and eukaryotic IDPs/IDRs.

Peptides from Conus Venoms which Affect Ca++ Entry into Neurons

AbstractThe venoms of fish-hunting Conus contain paralytic conotoxins and an unprecedented variety of other biologically-active peptides. Particularly noteworthy are peptides which inhibit calcium entry into neurons, the conantokins and w-conotoxins which target NMDA receptors and voltage-sensitive Ca channels respectively.The conantokins contain at least four residues of γ-carboxyglutamate (Gla), a post-translationally modified amino acid. Conantokins are folded into an α-helical conformation and bind acidic membranes in a calcium-dependent manner. Upon envenomation, conantokins may concentrate on appropriate membranes and target to peripheral fish NMDA receptors located on such membranes.The w-conotoxins are disulfide bonded peptides; a comparison of several w-conotoxin sequences reveals considerable sequence variability. Recent studies with w-conotoxins have also revealed a developmental specificity in mice. The results suggest that elements controlling breathing in neonatal mammals, but not in adults,...

Scaled quantum mechanical force field for cis- and trans-glycine in acidic solution

Abstract We obtain one set of ab initio scale factors for two conformational isomers of glycine in acidic solution. Ab initio calculations are performed using the 4-31 G basis set. Force constants from two fully optimized glycine-H + ·4H 2 O supermolecules are scaled using vibrational frequencies of glycine in water at pH 0.8. Similar calculations are shown for methylamine-H + and acetic acid. Reasonable agreement is obtained between calculated and experimental frequencies for glycine-H + , acetic acid, and methylamine-H + . Comparisons between scale factors for these molecules show that scale factors for the same symmetry coordinates are similar in most cases and indicate the extent to which scale factors can be transferred between these molecules. These scale factors are being used to build a scaled quantum mechanical force field for alanine, and subsequently for peptides in aqueous solutions.

Lipid-dependent Structural Changes of an Amphomorphic Membrane Protein

optical diffraction analysis supports this interpretation inthat thefiltered imagesarealmost identical to thosegivenby the particle lattice replicated from the external side.Thesmooth areas that give diffraction patterns (2 in Fig.4), which lie only slightly higher above the glass than theparticle lattice, representthesmoothlattice thatweidenti-fied in our previous paper as the true external fractureface. Weinterpretthis asalipid-carbon chainsurface. Thesmooth surfaces

Scaled quantum mechanical force field for glycine in basic solution

We obtain scale factors for three glycinate-nH2O ab initio force fields, using the 4–31G basis set, that can be used in building a scaled quantum mechanical force field for alanine and, subsequently, for peptides in aqueous solutions. Force constants from the fully optimized glycinate-nH2O supermolecules were scaled by using experimentally determined vibrational frequencies of glycine in water at pH 13. Similar calculations were performed for methylamine and acetate. Scale factors for the stretching modes of acetate are within 2% of the related scale factors for glycinate. The scale factor for the NH2 scissor mode in methylamine is also in agreement with that of glycinate. Changes in the scale factors as a function of the number of hydrating water molecules were also similar between glycinate and acetate. Amine groups showed relatively small changes. Scale factors for glycinate with no hydrating molecules were extrapolated from the supermolecule results, since the optimized structure of isolated glycinate obtained with the 4–31G basis set yielded one imaginary frequency. Good agreements between calculated and experimental frequencies for glycinate, acetate, and methyl amine were obtained for each set of scale factors. Scaling appears to compensate for the systematic effects of hydration on force constants, making it possible to obtain reliable frequency predictions for amino acids in water without resorting to expensive super-molecule calculations.

Effects of hydration on scale factors for ab initio force constantsII

Abstract Experimentally measured vibrational frequencies from the polar groups of peptides in aqueous solutions do not agree with frequencies calculated from scaled quantum mechanical force fields (SQMFF) using differential scale factors developed for molecules in the vapor phase. Measured stretching frequencies for carbonyl groups are more than 50 wavenumbers lower than the calculated values. On the other hand, frequencies for non-polar groups calculated using these scale factors are relatively accurate. Our goal is to develop a SQMFF that yields accurate calculated frequencies for peptides in aqueous solutions. To achieve this goal, it has been necessary to obtain scale factors for smaller hydrated molecules that can be used as a starting point for calculations on peptides. To this end, we have calculated scale factors for ab initio force constants for methylamine and protonated methylamine using a least-squares fit of calculated and experimental frequencies. We present a comparison of the experimental and calculated frequencies, along with their potential energy distributions, for both vapor and aqueous phases. We compare the scale factors derived from our measurements with changes observed in the ab initio force constants calculated for these molecules at various states of hydration. These force constants are calculated using fully optimized geometries for these hydrated molecules using the 4-31G, 4-31G**, 6-31G+, and 6–311G ∗∗ basis sets. The results here are similar in consistency with those found in our previous calculations on formic acid, acetic acid and acetone. The differences in scale factors between vapor and aqueous phase molecules are smaller than those previously found for polar groups, indicating that methylamine exhibits relatively non-polar behavior in solution with respect to its vibrational spectrum.

Circular dichroism and laser Raman spectroscopic analysis of the secondary structure ofCerebratulus lacteus toxin B-IV

The secondary structure ofCerebratulus lacteus toxin B-IV, a neurotoxic polypeptide containing 55 amino acid residues and four disulfide bonds, was experimentally estimated by computer analyses of toxin circular dichroism (CD) and laser Raman spectra. The CD spectrum of the toxin displayed typical α-helical peaks at 191, 208, and 222 nm. At neutralpH, the α-helix estimates from CD varied between 49 and 55%, when nonrepresentative spectrum analytical methods were used. Analysis of the laser Raman spectrum obtained at a much higher toxin concentration yielded a 78% α-helix estimate. Both CD and Raman spectroscopic methods failed to detect any β-sheet structure. The spectroscopic analyses revealed significantly more α-helix and less β-sheet for toxin B-IV than was predicted from its sequence. To account for the difference between the 49–55% helix estimate from CD spectra and the 78% helix estimate from the Raman spectrum, we postulate that some terminal residues are unfolded at the low toxin concentrations used for CD measurements but form helix at the high toxin concentration used for Raman measurements. Our CD observations showing thatCerebatulus toxin B-IV helix content increases about 15% in trifluoroethanol or at highpH are consistent with this interpretation.