Using artificially generated spectral data to improve protein secondary structure prediction from Fourier transform infrared spectra of proteins

Secondary structures of proteins have been predicted using neural networks from their Fourier transform infrared spectra. To improve the generalization ability of the neural networks, the training data set has been artificially increased by linear interpolation. The leave-one-out approach has been used to demonstrate the applicability of the method. Bayesian regularization has been used to train the neural networks and the predictions have been further improved by the maximum-likelihood estimation method. The networks have been tested and standard error of prediction (SEP) of 4.19% for alpha helix, 3.49% for beta sheet, and 3.15% for turns have been achieved. The results indicate that there is a significant decrease in the SEP for each type of structure parameter compared to previous works.     

Protein secondary structures in water from second-derivative amide I infrared spectra

Infrared spectra have been obtained for 12 globular proteins in aqueous solution at 20 degrees C. The proteins studied, which vary widely in the relative amounts of different secondary structures present, include myoglobin, hemoglobin, immunoglobulin G, concanavalin A, lysozyme, cytochrome c, alpha-chymotrypsin, trypsin, ribonuclease A, alcohol dehydrogenase, beta 2-microglobulin, and human class I major histocompatibility complex antigen A2. Criteria for evaluating how successfully the spectra due to liquid and gaseous water are subtracted from the observed spectrum in the amide I region were developed. Comparisons of second-derivative amide I spectra with available crystal structure data provide both qualitative and quantitative support for assignments of infrared bands to secondary structures. Band frequency assignments assigned to alpha-helix, beta-sheet, unordered, and turn structures are highly consistent among all proteins and agree closely with predictions from theory. alpha-Helix and unordered structures can each be assigned to only one band whereas multiple bands are associated with beta-sheets and turns. These findings demonstrate a method of analysis of second-derivative amide I spectra whereby the frequencies of bands due to different secondary structures can be obtained. Furthermore, the band intensities obtained provide a useful method for estimating the relative amounts of different structures.     

Fourier transform infrared analysis of bacteriorhodopsin secondary structure.

Fourier transform infrared spectroscopy at a resolution of 1 cm-1 has been used to study the conformation of dark-adapted bacteriorhodopsin in the native purple membrane, in H2O and D2O suspensions. A detailed analysis of the amide I bands was made using derivative and deconvolution techniques. Curve-fitting results of four independent experiments indicate, after estimation of the methodological errors, that native bacteriorhodopsin contains 52-73% alpha-helices, 13-19% reverse turns, 11-16% beta-sheets, and 3-7% unordered segments. Our analysis has enabled the identification of several components corresponding to alpha-helices, beta-sheets, and reverse turns. Besides the alpha I- and alpha II-helices (peaking at 1658 and 1665 cm-1), we propose that two more infrared bands arise from alpha-helical structures: one at 1650 cm-1 from alpha I and another one at 1642 cm-1 in H2O suspension, which could originate from type III beta-turns (i.e., one turn of 3(10)-helix). The relatively high content of reverse turns suggests the presence of one reverse turn per loop, plus another one in the C-terminal segment. On the other hand, several reasons argue that the calculated mean beta-sheet content of around 14% should be decreased somewhat. These beta-sheets could be located in the noncytoplasmatic links of the bacteriorhodopsin molecule.     

Fourier transform infrared spectroscopic analysis of protein secondary structures

Infrared spectroscopy is one of the oldest and well established experimental techniques for the analysis of secondary structure of polypeptides and proteins. It is convenient, non-destructive, requires less sample preparation, and can be used under a wide variety of conditions. This review introduces the recent developments in Fourier transform infrared （FTIR） spectroscopy technique and its applications to protein structural studies. The experimental skills, data analysis, and correlations between the FTIR spectroscopic bands and protein secondary structure components are discussed. The applications of FTIR to the second- ary structure analysis, conformational changes, structural dynamics and stability studies of proteins are also discussed.     

Protein secondary structure from Fourier transform infrared spectroscopy: a data base analysis

An infrared (ir) method to determine the secondary structure of proteins in solution using the amide I region of the spectrum has been devised. The method is based on the circular dichroism (CD) matrix method for secondary structure analysis given by Compton and Johnson (L. A. Compton and W. C. Johnson, 1986, Anal. Biochem. 155, 155-167). The infrared data matrix was constructed from the normalized Fourier transform infrared spectra from 1700 to 1600 cm-1 of 17 commercially available proteins. The secondary structure matrix was constructed from the X-ray data of the seventeen proteins with secondary structure elements of helix, beta-sheet, beta-turn, and other (random). The CD and ir methods were compared by analyzing the proteins of the CD and ir databases as unknowns. Both methods produce similar results compared to structures obtained by X-ray crystallographic means with the CD slightly better for helix conformation, and the ir slightly better for beta-sheet. The relatively good ir analysis for concanavalin A and alpha-chymotrypsin indicate that the ir method is less affected by the presence of aromatic groups. The concentration of the protein and the cell path length need not be known for the ir analysis since the spectra can be normalized to the total ir intensity in the amide I region. The ir spectra for helix, beta-sheet, beta-turn, and other, as extracted from the data-base, agree with the literature band assignments. The ir data matrix and the inverse matrix necessary to analyze unknown proteins are presented.     

Protein structure by Fourier transform infrared spectroscopy: second derivative spectra

Second derivative Fourier transform infrared spectra of the proteins ribonuclease A, hemoglobin, and beta-lactoglobulin A (native and denatured) have been obtained in deuterium oxide solution from 1350 to 1800 cm-1. The relationship of the original spectra to their second derivatives is briefly discussed. In the second derivative spectra, clearly resolved peaks are observed which can be associated with the alpha-helix, beta-strands, and turns. No protein spectra with such resolution have heretofore been reported. Tentative assignments are proposed, and the observed peaks are related to the secondary structure of the proteins studied. The data appear to present the first direct spectroscopic evidence of turns in a native protein.     

Second-derivative Fourier transform infrared spectra of seaweed galactans

The Fourier transform infrared spectra of agar, agarose, -, -, and -carrageenan, and ofChondrus canaliculatus, Iridaea ciliata, I. membranacea, I. laminarioides andGracilaria chilensis polysaccharides were recorded in the 4000–400 cm^-1 region. The bands in the second derivative mode are sharper and more bands are resolved than in the normal spectra.Agar, agarose andG. chilensis phycocolloids exhibit diagnostic bands at 790 and 713 cm^-1. -, - and -carrageenans, and native carrageenan-type polysaccharides fromC. canaliculatus andIridaea species exhibit bands at around 1160, 1140, 1100, 1070, 1040, 1008, 610, and 580 cm^-1. Therefore, FT-IR spectroscopy in the second-derivative mode may be applied to differentiate between agar- and carrageenan-types seaweed galactans.     

New Fourier transform infrared based computational method for peptide secondary structure determination. I. Description of method

Fourier transform infrared (FTIR) experiments in dimethylsulfoxide, a solvent incapable of H donation, demonstrate that H --> D isotopic replacement on the amide side of peptide bonds involves modifications of both the position and intensity of the amide I band. The effect of the isotopic substitution is particularly significant in the 1710-1670 and 1670-1650 cm(-1) regions, which are generally associated with beta-turns and alpha-helices. This behavior, attributed to the existence of intramolecular H-bonds in the polypeptide chain, is directly correlated to the presence of different secondary structures. Utilizing the effects induced by isotopic substitution, a method for the quantitative determination of the percentage of intramolecular H-bonds and the correlated secondary structures is proposed. The method consists of three principal steps: resolution of the fine structure of the amide I band with the determination of the number and position of the different components; reconstruction of the experimentally measured amide I band as a combination of Gaussian and Lorentzian functions, centered on the wave numbers set by band-narrowing methods, through a curve-fitting program; and quantitative determination of the population of the H-bonded carbonyls and the correlated secondary structures by comparison of the integrated intensities pertaining to the components with homologous wave numbers before and after isotopic exchange. The method is tested on a synthetic fragment of proocytocin that was previously analyzed by NMR techniques using the same solvent systems.     

A rapid method to screen for cell-wall mutants using discriminant analysis of Fourier transform infrared spectra

We have developed a rapid method to screen large numbers of mutant plants for a broad range of cell wall phenotypes using Fourier transform infrared (FTIR) microspectroscopy of leaves. We established and validated a model that can discriminate between the leaves of wild-type and a previously defined set of cell-wall mutants of Arabidopsis . Exploratory principal component analysis indicated that mutants deficient in different cell-wall sugars can be distinguished from each other. Discrimination of cell-wall mutants from wild-type was independent of variability in starch content or additional unrelated mutations that might be present in a heavily mutagenised population. We then developed an analysis of FTIR spectra of leaves obtained from over 1000 mutagenised flax plants, and selected 59 plants whose spectral variation from wild-type was significantly out of the range of a wild-type population, determined by Mahalanobis distance. Cell wall sugars from the leaves of selected putative mutants were assayed by gas chromatography-mass spectrometry and 42 showed significant differences in neutral sugar composition. The FTIR spectra indicated that six of the remaining 17 plants have altered ester or protein content. We conclude that linear discriminant analysis of FTIR spectra is a robust method to identify a broad range of structural and architectural alterations in cell walls, appearing as a consequence of developmental regulation, environmental adaptation or genetic modification.     

Estimation of protein secondary structure from the laser Raman amide I spectrum

A new method for estimating protein secondary structure from the laser Raman spectrum has been developed whereby the amide I Raman band of a protein is analyzed directly as a linear combination of amide I bands of proteins whose secondary structures are known. For 14 proteins, analyzed by removing each one from the reference set and analyzing its structure in terms of the remaining proteins, the average correlation coefficients between the Raman and X-ray diffraction estimates of helix, beta-strand, turn, and undefined were 0.98, 0.98, 0.82 and 0.35, respectively. Significant correlations were also observed for distinctions between alpha-helix (0.98) and disordered helix (0.82), and between parallel (0.82) and antiparallel (0.97) beta-sheets. The average standard deviation of these Raman estimates from the X-ray values is less than 4%. In addition, a singular value analysis of 20 Raman amide I spectra indicates that there may be as many as nine significant independent pieces of information present in the amide I region.     

Monitoring redox-dependent contribution of lipids in Fourier transform infrared difference spectra of complex I from Escherichia coli

Biochemical and crystallographic studies have shown that phospholipids are essential for the integrity and activity of membrane proteins. In the study presented here, we use electrochemically induced Fourier transform infrared (FTIR) spectroscopy to demonstrate variations occurring upon the presence and absence of lipids in NADH:ubiquinone oxidoreductase (complex I) from Escherchia coli by following the C=O vibration of the lipid molecule. Complex I is activated in the presence of lipids. Interestingly, in electrochemically induced FTIR difference spectra of complex I from E. coli, a new signal at 1744/1730 cm(-1) appears after addition of E. coli polar lipids, concomitant with the oxidized or reduced form, respectively. Absorbance spectra of liposomes from mixed lipids at different pH values demonstrate shifts for the carbonyl vibration depending on the environment. On this basis we suggest that lipids, though not redox active themselves, contribute in reaction-induced FTIR difference spectra, if a change occurs in the direct environment of the lipid during the observed reaction or coupled processes.     

Insight into protein structure and protein-ligand recognition by Fourier transform infrared spectroscopy

An overview of the application of Fourier transform infrared spectroscopy for the analysis of the structure of proteins and protein-ligand recognition is given. The principle of the technique and of the spectra analysis is demonstrated. Spectral signal assignments to vibrational modes of the peptide chromophore, amino acid side chains, cofactors and metal ligands are summarized. Several examples for protein-ligand recognition are discussed. A particular focus is heme proteins and, as an example, studies of cytochrome P450 are reviewed. Fourier transform infrared spectroscopy in combination with the various techniques such as time-resolved and low-temperature methods, site-directed mutagenesis and isotope labeling is a helpful approach to studying protein-ligand recognition.     

Calculation and use of protein-derived conformation-related spectra for the estimate of the secondary structure of proteins from their infrared spectra

This paper probes the calculation of conformation-related basis spectra from infrared spectra (amide I′band) of reference proteins of known conformational composition and, with their aid, the computation of conformations from the amide I′ band of globular proteins using in both approaches a least-squares, curve-fitting computer program for the analysis of the spectra. The following results were obtained. The infrared basis spectra for the α-helix conformation, the β-(antiparallel-chain pleated sheet) conformation and the ρ-conformation were calculated and their physical reality was substantiated. The basis spectra were shown to be similar when the absorption contributions of the side chains of amino acids were either neglected or taken into account (uncorrected or corrected basis spectra). The mutual correlation of the basis spectra, quantified by the roots of the diagonal elements of the inverse matrix, was found to be low enough only for the β-conformation to allow a statistically reliable estimate of the β-conformation content of proteins. The comparison of the percentages of the β-conformation derived from x-ray structural analysis or calculated from infrared spectra showed the suitability of the basis spectra for the rough estimate of the β-conformation percentages of proteins. The results were not significantly different when using the uncorrected or corrected basis spectra.     

Fourier transform infrared study of proteins with parallel beta-chains

Deconvolved and second derivative Fourier transform infrared spectra of the proteins flavodoxin and triosephosphate isomerase have been obtained in the 1600 to 1700 cm-1 (amide I) region. To our knowledge these results provide the first experimental infrared data on proteins with parallel beta-chains. Characteristic absorption bands for the parallel beta-segments are observed at 1626-1639 cm-1 (strong) and close to 1675 cm-1 (weak). Previous theoretical studies based on hypothetical models with large, regular beta-sheets had suggested bands close to 1650 and 1666 cm-1. Our new assignments were confirmed by band area measurements, which yield conformational information in good agreement with results from X-ray diffraction data. The spectra were compared with corresponding spectra of concanavalin A and carboxypeptidase A. The first contains only antiparallel beta-segments, the second "mixed" beta-segments, with some strands lying antiparallel and others parallel. None of the observed amide I band frequencies assigned to parallel beta-chains occurs in the 1650 cm-1 region associated with helical segments.     

Determination of protein secondary structure using factor analysis of infrared spectra

A method is presented for determining the secondary structural composition of a protein in aqueous solution from its infrared spectrum. A factor analysis approach is used to analyze the infrared spectra of 18 proteins whose crystal structures are known from X-ray studies. Factor analysis followed by multiple linear regression identifies those eigenspectra that correlate with the variation in properties described by the calibration set. The properties of interest in this study are % alpha-helix, % beta-sheet, and % turns. In the analysis of an unknown, the factor loadings required to reproduce its spectrum are substituted in the regression equation for each property to predict its secondary structural composition. The accuracy of the method was determined by removing each standard, in turn, from the calibration set and using a calibration set generated from the remainder to predict its composition. By this method we obtain standard errors of prediction of 3.9% for alpha-helix, 8.3% for beta-sheet, and 6.6% for turns. The method may also be applied to the spectra of proteins in 2H2O. The method has important advantages over those currently in use for the quantitative analysis of the infrared spectra of proteins. Manipulation of the spectrum is kept to a minimum, no curve-fitting is necessary, and the several amide I band components need not be assigned.     

Evaluation of the information content in infrared spectra for protein secondary structure determination

Fourier-transform infrared spectroscopy is a method of choice for the experimental determination of protein secondary structure. Numerous approaches have been developed during the past 15 years. A critical parameter that has not been taken into account systematically is the selection of the wavenumbers used for building the mathematical models used for structure prediction. The high quality of the current Fourier-transform infrared spectrometers makes the absorbance at every single wavenumber a valid and almost noiseless type of information. We address here the question of the amount of independent information present in the infrared spectra of proteins for the prediction of the different secondary structure contents. It appears that, at most, the absorbance at three distinct frequencies of the spectra contain all the nonredundant information that can be related to one secondary structure content. The ascending stepwise method proposed here identifies the relevance of each wavenumber of the infrared spectrum for the prediction of a given secondary structure and yields a particularly simple method for computing the secondary structure content. Using the 50-protein database built beforehand to contain as little fold redundancy as possible, the standard error of prediction in cross-validation is 5.5% for the alpha-helix, 6.6% for the beta-sheet, and 3.4% for the beta-turn.     

Fourier transform IR and Fourier transform Raman spectroscopy studies of metallothionein-III: amide I band assignments and secondary structural comparison with metallothioneins-I and -II

The secondary structures of porcine brain Cu(4)Zn(3)-metallothionein (MT)-III and Cd(5)Zn(2)MT-I, Cd(5)Zn(2)MT-II, and Zn(7)MT-I from rabbit livers in the solid state are investigated by Fourier transform IR spectroscopy (FTIR) and Fourier transform Raman spectroscopy (FT-Raman). The Cu(4)Zn(3)MT-III contains 26-28% beta-turns and half-turns, 13-14% 3(10)-helices, 47-49% random coils, and 11-12% beta-extended chains. The structural comparison of porcine brain Cu(4)Zn(3)MT-III with rabbit liver Cd(5)Zn(2)MT-I (II) and Zn(7)MT-I shows that the contents of the random coil structure are obviously increased. The results indicate that the insert of an acidic hexapeptide in the alpha domain of Cu(4)Zn(3)MT-III possibly forms an alpha helix. However, because the bands assigned to the alpha-helix and random coil structures are overlapped in the spectra, the content of random coil structures in Cu(4)Zn(3)MT-III is therefore higher than those in Cd(5)Zn(2)MT-I, Cd(5)Zn(2)MT-II, and Zn(7)MT-I.