Inaccurate secondary structure predictions often indicate protein fold switching

Although most proteins conform to the classical one‐structure/one‐function paradigm, an increasing number of proteins with dual structures and functions have been discovered. In response to cellular stimuli, such proteins undergo structural changes sufficiently dramatic to remodel even their secondary structures and domain organization. This “fold‐switching” capability fosters protein multi‐functionality, enabling cells to establish tight control over various biochemical processes. Accurate predictions of fold‐switching proteins could both suggest underlying mechanisms for uncharacterized biological processes and reveal potential drug targets. Recently, we developed a prediction method for fold‐switching proteins using structure‐based thermodynamic calculations and discrepancies between predicted and experimentally determined protein secondary structure (Porter and Looger, Proc Natl Acad Sci U S A 2018; 115:5968–5973). Here we seek to leverage the negative information found in these secondary structure prediction discrepancies. To do this, we quantified secondary structure prediction accuracies of 192 known fold‐switching regions (FSRs) within solved protein structures found in the Protein Data Bank (PDB). We find that the secondary structure prediction accuracies for these FSRs vary widely. Inaccurate secondary structure predictions are strongly associated with fold‐switching proteins compared to equally long segments of non‐fold‐switching proteins selected at random. These inaccurate predictions are enriched in helix‐to‐strand and strand‐to‐coil discrepancies. Finally, we find that most proteins with inaccurate secondary structure predictions are underrepresented in the PDB compared with their alternatively folded cognates, suggesting that unequal representation of fold‐switching conformers within the PDB could be an important cause of inaccurate secondary structure predictions. These results demonstrate that inconsistent secondary structure predictions can serve as a useful preliminary marker of fold switching.     

Analysis of the circular dichroism spectrum of proteins using the convex constraint algorithm: a practical guide.

Due to the time scale of circular dichroism (CD) measurements, it is theoretically possible to deconvolute such a spectrum if the pure CD spectra differ significantly from one another. In the last decade several methods have been published aiming at obtaining the conformational weights, or percentages (which are the coefficients for a linear combination) of the so-called typical secondary structural elements making up the three-dimensional structure of proteins. Two methods that can be used to determine the secondary structures of proteins are described here. The first method, called LINCOMB, is a simple algorithm based on a least-squares fit with a set of reference spectra representing the known secondary structures and yielding an estimation of weights attributed to alpha-helix, beta-pleated sheet (mainly antiparallel), beta-turns, unordered form, and aromatic/disulfide (or nonpeptide) contributions of the protein being analyzed. This method requires a "template" or reference curve set, which was obtained from the second method. The second method, "convex constraint analysis," is a general deconvolution method for a CD spectra set of any variety of conformational type. The algorithm, based on a set of three constraints, is able to deconvolute a set of CD curves to its common "pure"-component curves and conformational weights. To analyze a single CD spectrum with this method, the spectrum is appended to the data set used as a reference data set. As a way to determine the reliability of the algorithm and provide a guideline to its usage, some applications are presented.     

On the analysis of membrane protein circular dichroism spectra

Analysis of circular dichroism spectra of proteins provides information about protein secondary structure. Analytical methods developed for such an analysis use structures and spectra of a set of reference proteins. The reference protein sets currently in use include soluble proteins with a wide range of secondary structures, and perform quite well in analyzing CD spectra of soluble proteins. The utility of soluble protein reference sets in analyzing membrane protein CD spectra, however, has been questioned in a recent study that found current reference protein sets to be inadequate for analyzing membrane proteins. We have examined the performance of reference protein sets available in the CDPro software package for analyzing CD spectra of 13 membrane proteins with available crystal structures. Our results indicate that the reference protein sets currently available for CD analysis perform reasonably well in analyzing membrane protein CD spectra, with performance indices comparable to those for soluble proteins. Soluble + membrane protein reference sets, which were constructed by combining membrane proteins with soluble protein reference sets, gave improved performance in both soluble and membrane protein CD analysis.     

Neuro-fuzzy structural classification of proteins for improved protein secondary structure prediction

Fourier transform infrared (FTIR) spectroscopy is a very flexible technique for characterization of protein secondary structure. Measurements can be carried out rapidly in a number of different environments based on only small quantities of proteins. For this technique to become more widely used for protein secondary structure characterization, however, further developments in methods to accurately quantify protein secondary structure are necessary. Here we propose a structural classification of proteins (SCOP) class specialized neural networks architecture combining an adaptive neuro-fuzzy inference system (ANFIS) with SCOP class specialized backpropagation neural networks for improved protein secondary structure prediction. Our study shows that proteins can be accurately classified into two main classes "all alpha proteins" and "all beta proteins" merely based on the amide I band maximum position of their FTIR spectra. ANFIS is employed to perform the classification task to demonstrate the potential of this architecture with moderately complex problems. Based on studies using a reference set of 17 proteins and an evaluation set of 4 proteins, improved predictions were achieved compared to a conventional neural network approach, where structure specialized neural networks are trained based on protein spectra of both "all alpha" and "all beta" proteins. The standard errors of prediction (SEPs) in % structure were improved by 4.05% for helix structure, by 5.91% for sheet structure, by 2.68% for turn structure, and by 2.15% for bend structure. For other structure, an increase of SEP by 2.43% was observed. Those results were confirmed by a "leave-one-out" run with the combined set of 21 FTIR spectra of proteins.     

Analysis of protein circular dichroism spectra for secondary structure using a simple matrix multiplication

Inverse circular dichroism (CD) spectra are presented for each of the five major secondary structures of proteins: alpha-helix, antiparallel and parallel beta-sheet, beta-turn, and other (random) structures. The fraction of the each secondary structure in a protein is predicted by forming the dot product of the corresponding inverse CD spectrum, expressed as a vector, with the CD spectrum of the protein digitized in the same way. We show how this method is based on the construction of the generalized inverse from the singular value decomposition of a set of CD spectra corresponding to proteins whose secondary structures are known from X-ray crystallography. These inverse spectra compute secondary structure directly from protein CD spectra without resorting to least-squares fitting and standard matrix inversion techniques. In addition, spectra corresponding to the individual secondary structures, analogous to the CD spectra of synthetic polypeptides, are generated from the five most significant CD eigenvectors.     

Estimation of protein secondary structure from circular dichroism spectra: inclusion of denatured proteins with native proteins in the analysis

We have expanded our reference set of proteins used in the estimation of protein secondary structure by CD spectroscopy from 29 to 37 proteins by including 3 additional globular proteins with known X-ray structure and 5 denatured proteins. We have also modified the self-consistent method for analyzing protein CD spectra, SELCON3, by including a new selection criterion developed by W. C. Johnson, Jr. (Proteins Struct. Funct. Genet. 35, 307-312, 1999). The secondary structure corresponding to the denatured proteins was approximated to be 90% unordered, owing to the spectral similarity of the denatured proteins and unordered structures. We examined the thermal denaturation of ribonuclease T1 by CD using both the original and expanded sets of reference proteins and obtained more consistent results with the expanded set. The expanded set of reference proteins will be helpful for the determination of protein secondary structure from protein CD spectra with higher reliability, especially of proteins with significant unordered structure content and/or in the course of denaturation.     

Accuracy of protein secondary structure determination from circular dichroism spectra based on immunoglobulin examples

Strong contribution of the aromatic amino acid side chain chromophores to the far-UV circular dichroism (CD) spectra substantially distorts a relatively weak CD signal originating from beta sheet, the main type of immunoglobulin secondary structure. In this study we compared the secondary structure calculated from the far-UV CD spectra with the X-ray data for three antibody Fab fragments. Calculations were performed with three different algorithms, using two sets of reference proteins. Low standard deviations between all six estimates indicate stable mathematical solutions. Despite pronounced differences in the shape and amplitude of the CD spectra, we found a strong correlation between CD and X-ray data in the secondary structure for every protein studied. The number and average length of the secondary structure elements estimated from the CD spectra closely resemble those of the X-ray data. Agreement between spectroscopic and crystallographic results demonstrates that modern methods of secondary structure calculation are resilient to distortions of the far-UV CD spectra of immunoglobulins caused by aromatic side chain chromophores.     

The circular dichroism of tumor necrosis factor-α: Measurement into the vacuum UV and analysis for secondary structure

The circular dichroism (CD) spectrum of tumor necrosis factor-α has been measured into the vacuum UV to 168 nm. Analysis of the CD for secondary structure is in good agreement with X-ray diffraction results, but the analysis is somewhat unstable. Adding the CD of this protein together with its X-ray determined secondary structure to the basis set should improve subsequent analyses of CD spectra for other all-β proteins.     

Distinct circular dichroism spectroscopic signatures of polyproline II and unordered secondary structures: Applications in secondary structure analyses

Circular dichroism (CD) spectroscopy is a valuable method for defining canonical secondary structure contents of proteins based on empirically‐defined spectroscopic signatures derived from proteins with known three‐dimensional structures. Many proteins identified as being “Intrinsically Disordered Proteins” have a significant amount of their structure that is neither sheet, helix, nor turn; this type of structure is often classified by CD as “other”, “random coil”, “unordered”, or “disordered”. However the “other” category can also include polyproline II (PPII)‐type structures, whose spectral properties have not been well‐distinguished from those of unordered structures. In this study, synchrotron radiation circular dichroism spectroscopy was used to investigate the spectral properties of collagen and polyproline, which both contain PPII‐type structures. Their native spectra were compared as representatives of PPII structures. In addition, their spectra before and after treatment with various conditions to produce unfolded or denatured structures were also compared, with the aim of defining the differences between CD spectra of PPII and disordered structures. We conclude that the spectral features of collagen are more appropriate than those of polyproline for use as the representative spectrum for PPII structures present in typical amino acid‐containing proteins, and that the single most characteristic spectroscopic feature distinguishing a PPII structure from a disordered structure is the presence of a positive peak around 220nm in the former but not in the latter. These spectra are now available for inclusion in new reference data sets used for CD analyses of the secondary structures of soluble proteins.     

Molecular Dynamics Ensemble Refinement of Intrinsically Disordered Peptides According to Deconvoluted Spectra from Circular Dichroism

We have developed a computational method of atomistically refining the structural ensemble of intrinsically disordered peptides (IDPs) facilitated by experimental measurements using circular dichroism spectroscopy (CD). A major challenge surrounding this approach stems from the deconvolution of experimental CD spectra into secondary structure features of the IDP ensemble. Currently available algorithms for CD deconvolution were designed to analyze the spectra of proteins with stable secondary structures. Herein, our work aims to minimize any bias from the peptide deconvolution analysis by implementing a non-negative linear least-squares fitting algorithm in conjunction with a CD reference data set that contains soluble and denatured proteins (SDP48). The non-negative linear least-squares method yields the best results for deconvolution of proteins with higher disordered content than currently available methods, according to a validation analysis of a set of protein spectra with Protein Data Bank entries. We subsequently used this analysis to deconvolute our experimental CD data to refine our computational model of the peptide secondary structure ensemble produced by all-atom molecular dynamics simulations with implicit solvent. We applied this approach to determine the ensemble structures of a set of short IDPs, that mimic the calmodulin binding domain of calcium/calmodulin-dependent protein kinase II and its 1-amino-acid and 3-amino-acid mutants. Our study offers a, to our knowledge, novel way to solve the ensemble secondary structures of IDPs in solution, which is important to advance the understanding of their roles in regulating signaling pathways through the formation of complexes with multiple partners.     

Estimation of protein secondary structure from circular dichroism spectra: comparison of CONTIN, SELCON, and CDSSTR methods with an expanded reference set

We have expanded the reference set of proteins used in SELCON3 by including 11 additional proteins (selected from the reference sets of Yang and co-workers and Keiderling and co-workers). Depending on the wavelength range and whether or not denatured proteins are included in the reference set, five reference sets were constructed with the number of reference proteins varying from 29 to 48. The performance of three popular methods for estimating protein secondary structure fractions from CD spectra (implemented in software packages CONTIN, SELCON3, and CDSSTR) and a variant of CONTIN, CONTIN/LL, that incorporates the variable selection method in the locally linearized model in CONTIN, were examined using the five reference sets described here, and a 22-protein reference set. Secondary structure assignments from DSSP were used in the analysis. The performances of all three methods were comparable, in spite of the differences in the algorithms used in the three software packages. While CDSSTR performed the best with a smaller reference set and larger wavelength range, and CONTIN/LL performed the best with a larger reference set and smaller wavelength range, the performances for individual secondary structures were mixed. Analyzing protein CD spectra using all three methods should improve the reliability of predicted secondary structural fractions. The three programs are provided in CDPro software package and have been modified for easier use with the different reference sets described in this paper. CDPro software is available at the website: http://lamar.colostate.edu/ approximately sreeram/CDPro.     

Bioinformatics analyses of circular dichroism protein reference databases

MOTIVATION: Circular dichroism (CD) spectroscopy has become established as a key method for determining the secondary structure contents of proteins which has had a significant impact on molecular biology. Many excellent mathematical protocols have been developed for this purpose and their quality is above question. However, reference database sets of proteins, with CD spectra matched to secondary structure components derived from X-ray structures, provide the key resource for this task. These databases were created many years ago, before most CD spectrophotometers became standardized and before it was commonplace to validate X-ray structures prior to publication. The analyses presented here were undertaken to investigate the overall quality of these reference databases in light of their extensive usage in determining protein secondary structure content from CD spectra. RESULTS: The analyses show that there are a number of significant problems associated with the CD reference database sets in current use. There are disparities between CD spectra for the same protein collected by different groups. These include differences in magnitudes, peak positions or both. However, many current reference sets are now amalgamations of spectra from these groups, introducing inconsistencies that can lead to inaccuracies in the determination of secondary structure components from the CD spectra. A number of the X-ray structures used fall short on the validation criteria now employed as standard for structure determination. Many have substantial percentages of residues in the disallowed regions of the Ramachandran plot. Hence their calculated secondary structure components, used as a foundation for the reference databases, are likely to be in error. Additionally, the coverage of secondary structure space in the reference datasets is poorly correlated to the secondary structure components found in the Protein Data Bank. A conclusion is that a new reference CD database with cross-correlated, machine-independent CD spectra and validated X-ray structures that cover more secondary structure components, including diverse protein folds, is now needed. However, that reasonably accurate values for the secondary structure content of proteins can be determined from spectra is a testament to CD spectroscopy being a very powerful technique.     

A CD determination of the alpha-helix contents of the coat proteins of four filamentous bacteriophages: fd, IKe, Pf1, and Pf3

The CD spectra of four filamentous bacteriophages--fd, IKe, Pf1, and Pf3--were analyzed to determine the alpha-helix contents of their major coat proteins. Measured spectra included the 192-nm band so that analyses could be carried out over the full wavelength range of the reference spectra for protein secondary structures available (a) from globular proteins [J.T. Yang, C.S.C. Wu, and H.M. Martinez (1986) Methods in Enzymology 130, 208-269] and (b) from poly(L-lysine) [N. Greenfield and G.D. Fasman (1960) Biochemistry 8, 4108-4116]. Extended analyses were also performed with the addition of the spectrum of a model beta-turn to the Greenfield and Fasman reference set, with the spectrum of a short alpha-helix in the Yang et al. reference set, and with an estimate of the spectrum of Trp added to both reference sets. The reference set based on the simple poly(L-lysine) polypeptide, plus a spectrum of a model beta-turn or of Trp, gave reasonably good fits to the measured spectra for all four phages and yielded the largest percentages of alpha-helix. The class I phages--fd and IKe--had large percentages of alpha-helix of 98 +/- 2 and 97 +/- 5%, respectively, while the two class II phages--Pf1 and Pf3--had similar but smaller alpha-helix contents of 83 +/- 6 and 84 +/- 2, respectively. While these alpha-helix contents were within the ranges previously reported from CD spectra of these phages in solution, they were more precise, and they indicated that the coat proteins of the intact phages have CD spectra that are probably modeled better by the reference spectra of polypeptides than by those of globular proteins.     

Determination of secondary structure of globular proteins using circular dichroism spectra

A ridge regression method is presented for prediction of the secondary structure of proteins by the circular dichroism spectra (CD) from 190 to 236 nm. Eight types of the secondary structure were calculated on a microcalculator. The method is based on the X-ray data of Kabsh and Sander. The teaching rule is constructed on CD spectra of 30 proteins of all structural classes of the globular proteins (alpha, alpha/beta, alpha + beta and beta-proteins). The errors of the methods are analysed by removing each protein from the reference set and analyzing its structure in terms of the remaining proteins. Correlation coefficients and root-mean square deviations between CD and X-ray data were: 0.99 and 0.03 for alpha-helix, 0.86 and 0.02 for 3(10)-helix, 0.92 and 0.06 for antiparallel beta-sheet, 0.86 and 0.03 for parallel beta-sheet, 0.94 and 0.01 for T3 beta-turn, 0.85 and 0.02 for other beta-turn, 0.84 and 0.03 for S-bends, 0.83 and 0.04 for "random" structure.     

Deconvolution of the circular dichroism spectra of proteins: the circular dichroism spectra of the antiparallel beta-sheet in proteins.

A recently developed algorithm, called Convex Constraint Analysis (CCA), was successfully applied to determine the circular dichroism (CD) spectra of the pure beta-pleated sheet in globular proteins. On the basis of X-ray diffraction determined secondary structures, the original data set used (Perczel, A., Hollosi, M., Tusnady, G. Fasman, G.D. Convex constraint analysis: A natural deconvolution of circular dichroism curves of proteins, Prot. Eng., 4:669-679, 1991), was improved by the addition of proteins with high beta-pleated sheet content. The analysis yielded CD curves of the pure components of the main secondary structural elements (alpha-helix, antiparallel beta-pleated sheet, beta-turns, and unordered conformation), as well as a curve attributed to the "aromatic contribution" in the wavelength range of 195-240 nm. Upon deconvolution the curves obtained were assigned to various secondary structures. The calculated weights (percentages determining the contributions of each pure component curve in the measured CD spectra of a given protein) were correlated with the X-ray diffraction determined percentages in an assignment procedure and were evaluated. The Pearson product correlation coefficients (R) are significant for all five components. The new pure component curves, which were obtained through deconvolution of the protein CD spectra alone, are promising candidates for determining the percentages of the secondary structural components in globular proteins without the necessity of adopting an X-ray database. The CD spectrum of the CheY protein was interesting because it has the characteristic shape associated with the alpha-helical structure, but upon analysis yielded a considerable amount of beta-sheet in agreement with the X-ray structure.     

A holistic approach for protein secondary structure estimation from infrared spectra in H(2)O solutions

We present an improved technique for estimating protein secondary structure content from amide I and amide III band infrared spectra. This technique combines the superposition of reference spectra of pure secondary structure elements with simultaneous aromatic side chain, water vapor, and solvent background subtraction. Previous attempts to generate structural reference spectra from a basis set of reference protein spectra have had limited success because of inaccuracies arising from sequential background subtractions and spectral normalization, arbitrary spectral band truncation, and attempted resolution of spectroscopically degenerate structure classes. We eliminated these inaccuracies by defining a single mathematical function for protein spectra, permitting all subtractions, normalizations, and amide band deconvolution steps to be performed simultaneously using a single optimization algorithm. This approach circumvents many of the problems associated with the sequential nature of previous methods, especially with regard to removing the subjectivity involved in each processing step. A key element of this technique was the calculation of reference spectra for ordered helix, unordered helix, sheet, turns, and unordered structures from a basis set of spectra of well-characterized proteins. Structural reference spectra were generated in the amide I and amide III bands, both of which have been shown to be sensitive to protein secondary structure content. We accurately account for overlaps between amide and nonamide regions and allow different structure types to have different extinction coefficients. The agreement between our structure estimates, for proteins both inside and outside the basis set, and the corresponding determinations from X-ray crystallography is good.     

A holistic approach to protein secondary structure characterization using amide I band Raman spectroscopy

We have developed a holistic protein structure estimation technique using amide I band Raman spectroscopy. This technique combines the superposition of reference spectra for pure secondary structure elements with simultaneous aromatic, fluorescence, and solvent background subtraction, and is applicable to solution, suspension, and solid protein samples. A key component of this technique was the calculation of the reference spectra for ordered helix, unordered helix, and sheet, turns, and unordered structures from a series of well-characterized reference proteins. We accurately account for the overlap between the amide I and non-amide I regions and allow for different scattering efficiencies for different secondary structures. For hydrated samples, we allowed for the possibility that bound water spectra differ from the bulk water spectra. Our computed reference spectra compare well with previous experimental and theoretical results in the literature. We have demonstrated the use of these reference spectra for the estimation of secondary structures of proteins in solution, suspension, and dry solid forms. The agreement between our structure estimates and the corresponding determinations from X-ray crystallography is good.     

Extending CD spectra of proteins to 168 nm improves the analysis for secondary structures

The CD spectra for 10 proteins with known secondary structure have been extended from 178 to 168 nm. Combined with the data for 6 other proteins investigated previously, this produces a basis set of 16 proteins, which can be used to analyze CD spectra for secondary structure. Extending the spectra adds another CD band to the data and increases the information content from the equivalent of five to six. Analyzing the CD for each of the 16 proteins in the basis set with the 15 other proteins shows a modest improvement in the prediction of secondary structure with the extended CD spectra.     

Determination of the secondary structure of proteins from the amide I band of the laser Raman spectrum

The amide I band in the laser Raman spectrum of proteins has been resolved into six components, each representing residues in a different type of secondary structure. These structure types are ordered or bihydrogen-bonded helix (believed to be located in the center of helical segments), disordered or monohydrogen-bonded helix (believed to be located at the ends of helical segments), antiparallel beta sheet, parallel beta sheet, reverse turn, and undefined. The Raman spectrum representing 100% of each type of residue conformation has been computed from the solvent-subtracted Raman spectra of ten proteins with known secondary structure, plus poly-l-lysine using a least-squares solution of the overdetermined system of equations. Linear combinations of these reference spectra were then fitted to the experimental amide I spectra of these and other proteins to estimate the fractions of residues in these conformations. Statistical tests suggest that the discrimination between bihydrogen-bonded helix and monohydrogen-bonded helix is significant as is the discrimination between parallel and antiparallel β-sheet. However, the discrimination between random structure and turns has not yet been accomplished by these studies. The absolute difference between X-ray and Raman estimates of structure for 17 protein samples is generally less than 6%. We conclude that detailed and reasonably accurate estimates of secondary structure can be derived from the amide I spectra of proteins.     

Tchebichef image moment approach to the prediction of protein secondary structures based on circular dichroism

Circular dichroism (CD) spectroscopy is a widely used technique for the evaluation of protein secondary structures that has a significant impact for the understanding of molecular biology. However, the quantitative analysis of protein secondary structures based on CD spectra is still a hard work due to the serious overlap of the spectra corresponding to different structural motifs. Here, Tchebichef image moment (TM) approach is introduced for the first time, which can effectively extract the chemical features in CD spectra for the quantitative analysis of protein secondary structures. The proposed approach was applied to analyze reference set and the obtained results were evaluated by the strict statistical parameters such as correlation coefficient, cross‐validation correlation coefficient and root mean squared error. Compared with several specialized prediction methods, TM approach provided satisfactory results, especially for turns and unordered structures. Our study indicates that TM approach can be regarded as a feasible tool for the analysis of the secondary structures of proteins based on CD spectra. An available TMs package is provided and can be used directly for secondary structures prediction.