Prediction of beta-turns in proteins using neural networks

The use of neural networks to improve empirical secondary structure prediction is explored with regard to the identification of the position and conformational class of beta-turns, a four-residue chain reversal. Recently an algorithm was developed for beta-turn predictions based on the empirical approach of Chou and Fasman using different parameters for three classes (I, II and non-specific) of beta-turns. In this paper, using the same data, an alternative approach to derive an empirical prediction method is used based on neural networks which is a general learning algorithm extensively used in artificial intelligence. Thus the results of the two approaches can be compared. The most severe test of prediction accuracy is the percentage of turn predictions that are correct and the neural network gives an overall improvement from 20.6% to 26.0%. The proportion of correctly predicted residues is 71%, compared to a chance level of about 58%. Thus neural networks provide a method of obtaining more accurate predictions from empirical data than a simpler method of deriving propensities.     

An algorithm for protein secondary structure prediction based on class prediction

An algorithm has been developed to improve the success rate in the prediction of the secondary structure of proteins by taking into account the predicted class of the proteins. This method has been called the 'double prediction method' and consists of a first prediction of the secondary structure from a new algorithm which uses parameters of the type described by Chou and Fasman, and the prediction of the class of the proteins from their amino acid composition. These two independent predictions allow one to optimize the parameters calculated over the secondary structure database to provide the final prediction of secondary structure. This method has been tested on 59 proteins in the database (i.e. 10,322 residues) and yields 72% success in class prediction, 61.3% of residues correctly predicted for three states (helix, sheet and coil) and a good agreement between observed and predicted contents in secondary structure.     

Conformation of colipase: Prediction of the secondary structure,circular dichroism and 360 MHz proton NMR studies of porcine colipase A

The secondary structure of porcine colipase (93 residues) was established according to the predictive method of Chou and Fasman (Chou, P.Y. and Fasman, G.D. (1974) Biochemistry 13, 211–222 and 222–245). The relative composition of the conformational regions was as follows: 5% α-helix (region 39–44), 25% β-sheet (three regions, 7–11, 49–57 and 77–85) and eight β-turns corresponding to 32% of the polypeptide. Colipase contains a large proportion (about 35%) of unordered structure. Estimated values for the α-helix and β-sheet contents from the circular dichroism spectrum were in good accordance with the predicted model. A less satisfactory value was round for the β-turns. A characteristic feature of the far ultraviolet dichroic spectrum is the presence of an unusual positive band at 225 nm that might be indicative of a particular spatial arrangement of the chromophores in the molecule. Two tyrosines (Tyr₅₆ and Tyr₅₇) and one histidine (His₈₆) are at close vicinity in the three dimensional structure of the protein as shown by proton NMR studies. These residues are located at the end of two β-sheet hydrophobic regions (49–57 and 77–85) which might play a role in the association of colipase with the lipid-water interface as indicated by results of the NMR studies of the taurodeoxycholate-colipase complex.     

The structure of human acidic fibroblast growth factor and its interaction with heparin.

The secondary and tertiary structure of recombinant human acidic fibroblast growth factor (aFGF) has been characterized by a variety of spectroscopic methods. Native aFGF consists of ca. 55% beta-sheet, 20% turn, 10% alpha-helix, and 15% disordered polypeptide as determined by laser Raman, circular dichroism, and Fourier transform infrared spectroscopy; the experimentally determined secondary structure content is in agreement with that calculated by the semi-empirical methods of Chou and Fasman (Chou, P. Y., and Fasman, G. C., 1974, Biochemistry 13, 222-244) and Garnier et al. (Garnier, J. O., et al., 1978, J. Mol. Biol. 120, 97-120). Using the Garnier et al. algorithm, the major secondary structure components of aFGF have been assigned to specific regions of the polypeptide chain. The fluorescence spectrum of native aFGF is unusual in that it is dominated by tyrosine fluorescence despite the presence of a tryptophan residue in the protein. However, tryptophan fluorescence is resolved upon excitation above 295 nm. The degree of tyrosine and tryptophan solvent exposure has been assessed by a combination of ultraviolet absorption, laser Raman, and fluorescence spectroscopy; the results suggest that seven of the eight tyrosine residues are solvent exposed while the single tryptophan is partially inaccessible to solvent in native aFGF, consistent with recent crystallographic data. Denaturation of aFGF by extremes of temperature or pH leads to spectroscopically distinct conformational states in which contributions of tyrosine and tryptophan to the fluorescence spectrum of the protein vary. The protein is unstable at physiological temperatures. Addition of heparin or other sulfated polysaccharides does not affect the spectroscopic characteristics of native aFGF. These polymers do, however, dramatically stabilize the native protein against thermal and acid denaturation as determined by differential scanning calorimetry, circular dichroism, and fluorescence spectroscopy. The interaction of aFGF with such polyanions may play a role in controlling the activity of this growth factor in vivo.     

Prediction of β-sheets in immunoglobulin chains. Comparison of various methods and an expanded 20 × 20 table for evaluation of the effects of nearest-neighbors on conformations of middle amino acids in proteins

The extraordinarily large number of immunoglobulins renders them an intriguing class of molecules for attempts to predict their conformations. The predictive method applied, using a 20 × 20 table of the observed effects of nearest-neighboring amino acids on the conformation (Φ,Ψ angles) of the middle residue in known proteins, indicates positions of tri-peptides that tend to break α-helices or regular β-sheets. This 20 × 20 table is derived from data on 19 proteins, as compared with the earlier version based on 12 proteins, and includes a separate listing of residues of β-turns that have helical Φ,Ψ values. Secondary conformations predicted by methods of Chou and Fasman, Lim and Burgess, Ponnuswamy, and Scheraga have also been compared; for all three methods, wrong predicitons of residues in β-sheet conformation exceed correct ones. Better predictions are obtained when there is agreement with two or three of the methods. If there is consistent overprediction of β-structure, as with the Chou and Fasman method, the use of the β-sheet-breaking tripeptides can improve pre-dictability somewhat.     

Cholera toxin A subunit: functional sites correlated with regions of secondary structure

The A subunit of cholera toxin contains the ADP-ribosyltransferase activity in its major constituent polypeptide A1 (Mr 23,000) which is responsible for the elevation of cAMP typically observed with most mammalian cell types after exposure to the toxin. The primary structure of the A subunit, recently established by sequence analyses, is presented and used as the basis for the secondary structure prediction according to the method of Chou and Fasman. The results indicated the presence of 27% alpha-helix, 25% beta-structure, 12% beta-turn, and 36% random coil. The majority of the beta-structure consisted of six strands located in the NH2-terminal portion of the molecule (residues 33-106) covering one-half of the region corresponding to the A1 polypeptide portion. The beta-sheet domain led immediately into the active site region characterized by the alternating structures of beta-pleated sheet and alpha-helix (residues 95-140) similar to that reported for other NAD+ binding proteins. The presence of this structural feature in the region was confirmed by the use of another predictive method (J. Garnier et al., J. Mol. Biol. 1978, 120, 97-120). In addition, two regions (residues 14-18 and 200-214), previously identified to contain binding sites for the B subunit as evidenced by chemical modification and monoclonal antibody studies, were found to be in alpha-helix configuration.     

Structural and spectroscopic comparison of manganese-containing superoxide dismutases

Predicted secondary structures and optical properties of four manganese-containing superoxide dismutases isolated from Saccharomyces cerevisiae, Bacillus stearothermophilus, Escherichia coli and human liver are compared. The structural predictions are further compared with the known crystal structure of the manganese-containing superoxide dismutase from Thermus thermophilus HB8. The secondary structures of the four dismutases are predicted by the methods of Chou and Fasman (Adv. Enzymol. 47 (1978) 45-148), Garnier et al. (J. Mol. Biol. 120 (1978) 97-120) and Lim (J. Mol. Biol. 88 (1974) 873-894). The three models show satisfactory agreement and predict that the enzymes have a mixed alpha-helix and beta-sheet structure, and that they have homologous structures. The former conclusion is also reached from an analysis of the hydrophobic character of the amino-acid sequences of the four proteins according to Kyte and Doolittle (J. Mol. Biol. 157 (1982) 105-132). The calculation of the secondary structure based on the 185-260 nm circular dichroism spectrum of manganese-containing superoxide dismutase from S. cerevisiae reveals that the enzyme consists of 61% alpha-helix, 13% beta-sheet, 11% turn and 8% random coil conformations, which is in good accordance with the prediction based on the amino-acid sequences. Comparison of the 400-700 nm circular dichroism spectra of manganese-containing superoxide dismutase from S. cerevisiae, E. coli and T. thermophilus demonstrates that manganese atoms have homologous coordination in the three enzymes. This investigation based on primary structures and spectral properties indicates that the four dismutases have the same overall structure. Since the structural predictions are in good agreement with the structure found for the manganese-containing superoxide dismutase from T. thermophilus HB8, it can be concluded that this structure is representative for the four enzymes and probably for manganese-containing superoxide dismutases in general.     

Primary structure of Stoichactis helianthus cytolysin III

The primary structure of Stoichactis helianthus cytolysin III has been determined by automated Edman degradation of the intact protein and of peptides derived therefrom by hydrolysis with trypsin and staphylococcal protease and by chemical cleavage with cyanogen bromide and o-iodosobenzoic acid. As a result of these studies, the positions of all 153 amino acid residues of toxin III have been unambiguously determined. Most regions of sequence were determined two times in different types of digests of the protein. A number of highly hydrophobic regions of sequence, which may be functionally significant, have been identified, including a region rich in tyrosine and tryptophan (residues 86-98). The secondary structure of toxin III has been predicted by Chou-Fasman analysis (Chou, P.Y., and Fasman, G.D. (1978) Annu. Rev. Biochem. 47, 251-276) of the primary structure. The predicted secondary structure contains 16% alpha-helix and 31% beta-structure.     

The prediction of the conformation of membrane proteins from the sequence of amino acids.

The methods of Chou & Fasman [Biochemistry (1974) 13, 211-222, 222-245] and of Lim [J. Mol. Biol. (1974)88, 857-872, 873-894] for predicting secondary structure from amino acid sequence have been applied to five predominantly helical membrane-associated peptides. The predictions from the method of Lim (1974a,b) are consistent with the experimental observations, whereas those from Chou & Fasman (1974a,b), although not inconsistent with alpha-helix, favour a beta-structure for several very hydrophobic regions. The results may be rationalized in terms of the effect of the solvent on the conformation of a polypeptide.     

A simple method for predicting the secondary structure of globular proteins: implications and accuracy

A method is presented for predicting the secondary structureof globular proteins from their amino acid sequence. It is basedon a rigorous statistical exploitation of the well-known biologicalfact that the amino acid compositions of each secondary structureare different. We also propose an evaluation process that allowsus to estimate the capacity of a method to predict the secondarystructure of a new protein which does not have any homologousproteins whose structure is already known. This evaluation processshows that our method has a prediction accuracy of 58.7% overthree states for the 62 proteins of the Kabsch and Sander (1983a)data bank. This result is better than that obtained by the mostwidely used methods—Lim (1974), Chou and Fasman (1978)and Garnier et al. (1978)—and also than that obtainedby a recent method based on local homologies (Levin et al.,1986). Our prediction method is very simple and may be implementedon any microcomputer and even on programmable pocket calculators.A simple Pascal implementation of the method prediction algorithmis given. The interpretation of our results in terms of proteinfolding and directions for further work are discussed. Received on December 15, 1987; accepted on April 12, 1988     

Investigations of primary and secondary structure of porcine ubiquitin. Its N epsilon-acetylated lysine derivative

N epsilon-acetylation in vitro of internal lysyl residues of Ub by p-nitro-phenyl acetate at pH 8.0 was performed. The position of acetylation sites are determined. (e.g. Fully acetylated: Lys-6, Lys-11 and Lys-33; partially free internal lysines: Lys-27, Lys-29; Lys-48 and probably Lys-63.) 55 cycles Edman degradation were performed and the first 53 N-terminal residues were identified. Secondary structural studies of ubiquitin have been carried out using the circular dichroism (CD) technique. No changes are noted upon heating to 100 degrees C at neutral pH even in the presence of 8 M urea but in 6 M guanidine-HCl extensive modification results. Ubiquitin with an average of 4.4 of its 7 lysines in the N epsilon-acetyl form shows little deviation from native protein. After reduction with dithiothreitol and subsequent removal of the mercaptan, significant changes in the secondary structure are noted. Circular dichroic measurements of ubiquitin indicated an alpha-helical content of about 10% whereas the secondary structural predictions of Chou and Fasman suggest a level of about 45%.     

Amino acid sequence of the biotinyl subunit from transcarboxylase.

The complete amino acid sequence of the biotinyl subunit from the enzyme transcarboxylase of Propionibacterium shermanii has been determined from the structures of overlapping tryptic and cyanogen bromide peptides together with sequenator analysis on the whole subunit. The subunit contains 123 amino acid residues. Eleven of nineteen residues in the region of biotin attachment, when compared to pyruvate carboxylase from avian liver (Rylatt, D. B., Keech, D. B., and Wallace, J. C. (1977) Arch. Biochem. Biophys. 183, 113-122), were found to be in identical positions relative to biocytin. There was less homology with acetyl-CoA carboxylase from Escherichia coli (Sutton, M. R., Fall, R. R., Nervi, A. M., Alberts, A. W., Vagelos, P. R., and Bradshaw, R. A. (1977) J. Biol. Chem. 252, 3934-3940), but in all of these biotin enzymes there was an alanylmethionyl-biocytinyl-methionine sequence. The secondary structure of the biotinyl subunit has been estimated using the method of Chou and Fasman (Chou, P. Y., and Fasman, G. D. (1978) Adv. Enzymol. 47, 45-148) and considered in relationship to the role of the biotinyl subunit in the structure and function in transcarboxylase.     

Amino acid propensities for secondary structures are influenced by the protein structural class

Amino acid propensities for secondary structures were used since the 1970s, when Chou and Fasman evaluated them within datasets of few tens of proteins and developed a method to predict secondary structure of proteins, still in use despite prediction methods having evolved to very different approaches and higher reliability. Propensity for secondary structures represents an intrinsic property of amino acid, and it is used for generating new algorithms and prediction methods, therefore our work has been aimed to investigate what is the best protein dataset to evaluate the amino acid propensities, either larger but not homogeneous or smaller but homogeneous sets, i.e., all-alpha, all-beta, alpha-beta proteins. As a first analysis, we evaluated amino acid propensities for helix, beta-strand, and coil in more than 2000 proteins from the PDBselect dataset. With these propensities, secondary structure predictions performed with a method very similar to that of Chou and Fasman gave us results better than the original one, based on propensities derived from the few tens of X-ray protein structures available in the 1970s. In a refined analysis, we subdivided the PDBselect dataset of proteins in three secondary structural classes, i.e., all-alpha, all-beta, and alpha-beta proteins. For each class, the amino acid propensities for helix, beta-strand, and coil have been calculated and used to predict secondary structure elements for proteins belonging to the same class by using resubstitution and jackknife tests. This second round of predictions further improved the results of the first round. Therefore, amino acid propensities for secondary structures became more reliable depending on the degree of homogeneity of the protein dataset used to evaluate them. Indeed, our results indicate also that all algorithms using propensities for secondary structure can be still improved to obtain better predictive results.     

Predicted secondary structures of four penicillin beta-lactamases and a comparison with two lysozymes.

We have predicted the secondary structures of four beta-lactamases (Bacillus cereus, Bacillus licheniformis, Staphylococcus aureus, and Escherichia coli R-TEM) by the statistical method of Chou & Fasman as well as by the information theory method of Garnier et al. The secondary structures of all four beta-lactamases are of the alpha/beta type (Levitt & Chothia's nomenclature), with helices at N- and C-termini. There are about eight short regions each of alpha-helical (30--50%) and beta-strand (10--20%) structure separated by about 20 reverse turns. The conformation of the Gram-positive and Gram-negative beta-lactamases are generally similar although a few differences are predicted between the S.aureus and E.coli structures. Surprisingly, the two bacilli structures differ significantly in three short regions. In all four enzymes the region near the catalytically-implicated tyrosine has similar secondary structure. The secondary structure of hen egg white lysozyme, a penicillin-binding enzyme, as well as T4 phage lysozyme, has similarities to the N-terminal half of the penicillin-destroying beta-lactamases.     

Chicken liver H-protein, a component of the glycine cleavage system. Amino acid sequence and identification of the N epsilon-lipoyllysine residue

The complete amino acid sequence of H-protein from chicken liver was determined by aligning peptides obtained by cyanogen bromide, endoproteinase Lys-C, Staphylococcus aureus V8 protease, and chymotrypsin cleavage together with the partial NH2- and COOH-terminal sequence of the intact protein. H-protein consists of 125 amino acids and a lipoic acid moiety linked to lysine 59. The sequence is: (sequence in text). The lysyl residue involved in lipoic acid attachment is indicated with an asterisk. The molecular weight including lipoic acid is calculated to be 13,883. From the secondary structure predicted by the method of Chou and Fasman (Chou, P. Y., and Fasman, G. D. (1978) Adv. Enzymol. 47, 45-148) the lipoic acid binding region shows alpha-helical structure and is predicted to be an interior portion of the protein from the hydropathic profile according to Kyte and Doolittle (Kyte, J., and Doolittle, R. F. (1982) J. Mol. Biol. 157, 105-132).     

Secondary structure of the variant surface glycoproteins of trypanosomes

The secondary structure of seven variant surface glycoproteins (VSGs) of trypanosomes has been determined by Raman spectroscopy. They are all predominantly alpha-helical, the alpha-helix content varying between 50 and 60%. The beta-strand content varies between 20 and 25%, and the content of beta-turn and nonregular structures is about 25%. For three VSGs the N-terminal domain obtained by proteolytic cleavage was found to have essentially the same secondary structure as the complete VSGs. For three VSGs a secondary structure prediction has been performed applying the rules of Chou and Fasman. In all cases, two long alpha-helices extending over about 50 residues or 80 A are predicted in agreement with the X-ray diffraction data of Freymann et al. [(1984) Nature 311, 167-169] and Metcalf et al. [(1987) Nature 325, 84-86]. The region between the two alpha-helical segments exhibits a high potential of beta-turns, suggesting that this segment may be exposed on the cell surface and carry major antigenic determinants.     

Structure of secretory protein IV from rat seminal vesicles

The complete amino acid sequence of rat SVS-IV protein, consisting of 90 residues, has been determined. The sequence of rat SVS-IV protein is the first seminal vesicle secretory protein determined and it does not show any homology with other known protein sequences. The secondary structure of SVS-IV protein is analyzed by methods of Fasman and Chou indicates that this protein contains 53% alpha-helix, 36% beta-turn and 11% random coil.     

Conformational alterations in the proximal portion of the yeast invertase signal peptide do not block secretion

Summary Various amino acid insertions have been introduced into the proximal portion of the signal sequence of secreted yeast invertase. The altered invertase genes have been reintroduced into yeast and monitored for their ability to direct synthesis of secreted invertase in vivo. The insertions should alter the signal polypeptide local secondary structure as predicted by the Chou and Fasman rules (1978). Secretion of these altered invertase polypeptides is not blocked by the amino acid insertions.