Protein secondary structure prediction with SPARROW

A first step toward predicting the structure of a protein is to determine its secondary structure. The secondary structure information is generally used as starting point to solve protein crystal structures. In the present study, a machine learning approach based on a complete set of two-class scoring functions was used. Such functions discriminate between two specific structural classes or between a single specific class and the rest. The approach uses a hierarchical scheme of scoring functions and a neural network. The parameters are determined by optimizing the recall of learning data. Quality control is performed by predicting separate independent test data. A first set of scoring functions is trained to correlate the secondary structures of residues with profiles of sequence windows of width 15, centered at these residues. The sequence profiles are obtained by multiple sequence alignment with PSI-BLAST. A second set of scoring functions is trained to correlate the secondary structures of the center residues with the secondary structures of all other residues in the sequence windows used in the first step. Finally, a neural network is trained using the results from the second set of scoring functions as input to make a decision on the secondary structure class of the residue in the center of the sequence window. Here, we consider the three-class problem of helix, strand, and other secondary structures. The corresponding prediction scheme "SPARROW" was trained with the ASTRAL40 database, which contains protein domain structures with less than 40% sequence identity. The secondary structures were determined with DSSP. In a loose assignment, the helix class contains all DSSP helix types (α, 3-10, π), the strand class contains β-strand and β-bridge, and the third class contains the other structures. In a tight assignment, the helix and strand classes contain only α-helix and β-strand classes, respectively. A 10-fold cross validation showed less than 0.8% deviation in the fraction of correct structure assignments between true prediction and recall of data used for training. Using sequences of 140,000 residues as a test data set, 80.46% ± 0.35% of secondary structures are predicted correctly in the loose assignment, a prediction performance, which is very close to the best results in the field. Most applications are done with the loose assignment. However, the tight assignment yields 2.25% better prediction performance. With each individual prediction, we also provide a confidence measure providing the probability that the prediction is correct. The SPARROW software can be used and downloaded on the Web page http://agknapp.chemie.fu-berlin.de/sparrow/ .     

Interaction of a Rhodococcus sp. trehalose lipid biosurfactant with model proteins: thermodynamic and structural changes

One major application of surfactants is to prevent aggregation during various processes of protein manipulation. In this work, a bacterial trehalose lipid (TL) with biosurfactant activity, secreted by Rhodococcus sp., has been identified and purified. The interactions of this glycolipid with selected model proteins have been studied by using differential scanning calorimetry (DSC), Fourier-transform infrared (FTIR) spectroscopy, isothermal titration calorimetry (ITC), and fluorescence spectroscopy. Bovine serum albumin (BSA) and cytochrome c (Cyt-c) have been chosen because of their quite different secondary structures: BSA contains essentially no β-sheets and an average 66% α-helix, whereas Cyt-c possesses up to 25% β-sheets and up to 45% α-helical structure. Differential scanning calorimetry shows that addition of TL to BSA at concentrations below the critical micelle concentration (cmc) shifts the thermal unfolding temperature to higher values. FTIR indicates that TL does not alter the secondary structure of native BSA, but the presence of TL protects the protein toward thermal denaturation, mainly by avoiding formation of β-aggregates. Studies on the intrinsic Trp fluorescence of BSA show that addition of TL to the native protein results in conformational changes. BSA unfolding upon thermal denaturation in the absence of TL makes the Trp residues less accessible to the quencher, as shown by a decrease in the value of Stern-Volmer dynamic quenching constant, whereas denaturation in the presence of the biosurfactant prevents unfolding, in agreement with FTIR results. In the case of Cyt-c, interaction with TL gives rise to a new thermal denaturation transition, as observed by DSC, at temperatures below that of the native protein, therefore facilitating thermal unfolding. Binding of TL to native BSA and Cyt-c, as determined by ITC, suggests a rather nonspecific interaction of the biosurfactant with both proteins. FTIR indicates that TL slightly modifies the secondary structure of native Cyt-c, but protein denaturation in the presence of TL results in a higher proportion of β-aggregates than in its absence (20% vs 3.9%). The study of Trp fluorescence upon TL addition to Cyt-c results in a completely opposite scenario to that described above for BSA. In this case, addition of TL considerably increases the value of the dynamic quenching constant, both in native and denatured protein; that is, the interaction with the glycolipid induces conformational changes which facilitate the exposure of Trp residues to the quencher. Considering the structures of both proteins, it could be derived that the characteristics of TL interactions, either promoting or avoiding thermal unfolding, are highly dependent on the protein secondary structure. Our results also suggest the rather unspecific nature of these interactions. These might well involve protein hydrophobic domains which, being buried into the protein native structures, become exposed upon thermal unfolding.     

用概率统计法估算天花粉蛋白的二级结构

本文应用Levitt提出的蛋白质中氨基酸对二级结构的特定偏爱的统计方法估算天花粉蛋白的二级结构。计算结果表明,在天花粉蛋白中α-螺旋为32％,β-折叠为28％,反转为20％,其他类型为20％。这一计算结果与X射线衍射研究、圆二色性谱研究的结果相近。     

New β-strand templates constrained by Huisgen cycloaddition

New peptidic templates constrained into a β-strand geometry by linking acetylene and azide containing P(1) and P(3) residues of a tripeptide by Huisgen cycloaddition are presented. The conformations of the macrocycles are defined by NMR studies and those that best define a β-strand are shown to be potent inhibitors of the protease calpain. The β-strand templates presented and defined here are prepared under optimized conditions that should be suitable for targeting a range of proteases and other applications requiring such a geometry.     

Molecular insight for the role of key residues of calreticulin in its binding activities: A computational study

Calreticulin (CRT) is localized to and has functions in multiple cellular compartments, including the cell surface, the endoplasmic reticulum, and the extracellular matrix. Mutagenesis studies have identified several residues on a concave β-sheet surface of CRT critical for CRT binding to carbohydrate and other proteins/peptides. How the mutations of these key residues in CRT affect the conformation and dynamics of CRT, further influencing CRT binding to carbohydrates and other proteins to signal the important biological activities remain unknown. In this study, we investigated the effect of three key point mutations (C105A, C137A and W319A) on CRT conformation and dynamics via atomistic molecular dynamics simulations. Results show that these three key residues mutations induced the changes of CRT local backbone flexibility and secondary structure of CRT N-domain, which could further affect CRT’s binding activity. C137A mutation led to dramatic decrease of the overall size of CRT due to the P-domain fold back to the globular domain and formed new inter-domain contacts, which can cause blockage of CRT’s binding with other large substrates. Furthermore, for CRT concave β-strand surface patch containing lectin binding site, CRT C105A, C137A and W319A point mutation resulted in the changes in solvent accessible surface area, key residues’ side chain atom positions and dynamical correlated motions between residues. All these changes could directly affect CRT binding behavior. Results of this study provide molecular and structural insights into understanding the role of key residues of CRT in its binding behavior.     

Parallel and antiparallel β-strands differ in amino acid composition and availability of short constituent sequences

One of the important secondary structures in proteins is the β-strand. However, due to its complexity, it is less characterized than helical structures. Using the 1641 representative three-dimensional protein structure data from the Protein Data Bank, we characterized β-strand structures based on strand length and amino acid composition, focusing on differences between parallel and antiparallel β-strands. Antiparallel strands were more frequent and slightly longer than parallel strands. Overall, the majority of β-sheets were antiparallel sheets; however, mixed sheets were reasonably abundant, and parallel sheets were relatively rare. Notably, the nonpolar, aliphatic hydrocarbon amino acids, valine, isoleucine, and leucine were observed at a high frequency in both strands but were more abundant in parallel than in antiparallel strands. The relative amino acid occurrence in β-sheets, especially in parallel strands, was highly correlated with amino acid hydrophobicity. This correlation was not observed in α-helices and 3(10)-helices. In addition, we examined the frequency of 400 amino acid doublets and 8000 amino acid triplets in β-strands based on availability, a measurement of the relative counts of the doublets and triplets. We identified some triplets that were specifically found in either parallel or antiparallel strands. We further identified "zero-count triplets" which did not occur in either parallel or antiparallel strands, despite the fact that they were probabilistically supposed to occur several times. Taken together, the present study revealed essential features of β-strand structures and the differences between parallel and antiparallel β-strands, which can potentially be applied to the secondary structure prediction and the functional design of protein sequences in the future.     

Study on the Conformation of Cyclodecapeptide Loloatin C with Obvious Antibiotic Activity

The conformation of cyclodecapeptide loloatin C with obvious antibiotic activity has been investigated in 2,2,2-trifluoroethanol/sodium acetate buffer solution and then characterized by Fr-IR, CD and NMR spectrum. The results of FT-IR show that there exists β-strand or β-tum secondary structure in the molecule. According to the CD spectrum, the helical turn is dominant but the β-turn structure also exists. Conformation of the whole molecule is probably a helical β-turn.The chemical shifts and coupling constants prove the existence of a β-structure in the regions of Val,Orn2 and Leu3. NOESY data and temperature gradients of amide protons suggest that the molecular conformation is a dumbbell-like structure with the waist located between ornithyl (position 2) and D-phenylalanyl (position 7) and β-turn on both ends.     

β-Sheet Core of Tau Paired Helical Filaments Revealed by Solid-State NMR

One of the hallmarks of Alzheimer's disease is the self-assembly of the microtubule-associated protein tau into fibers termed "paired helical filaments" (PHFs). However, the structural basis of PHF assembly at atomic detail is largely unknown. Here, we applied solid-state nuclear magnetic resonance (ssNMR) spectroscopy to investigate in vitro assembled PHFs from a truncated three-repeat tau isoform (K19) that represents the core of PHFs. We found that the rigid core of the fibrils is formed by amino acids V306 to S324, only 18 out of 99 residues, and comprises three β-strands connected by two short kinks. The first β-strand is formed by the well-studied hexapeptide motif VQIVYK that is known to self-aggregate in a steric zipper arrangement. Results on mixed [(15)N:(13)C]-labeled K19 fibrils show that β-strands are stacked in a parallel, in-register manner. Disulfide bridges formed between C322 residues of different molecules lead to a disturbance of the β-sheet structure, and polymorphism in ssNMR spectra is observed. In particular, residues K321-S324 exhibit two sets of resonances. Experiments on K19 C322A PHFs further confirm the influence of disulfide bond formation on the core structure. Our structural data are supported by H/D exchange NMR measurements on K19 as well as a truncated four-repeat isoform of tau (K18). Site-directed mutagenesis studies show that single-point mutations within the three different β-strands result in a significant loss of PHF aggregation efficiency, highlighting the importance of the β-structure-rich regions for tau aggregation.     

Evidence for conformational differences in aqueous and crystalline structures of α-bungarotoxin and cobratoxin

Laser Raman spectroscopy has been employed to investigate the structures of α-bungarotoxin (Bungarus multicinctus) and cobratoxin (Naja naja siamensis) in H₂O and D₂O solutions. Structures of the aqueous neurotoxins are compared with one another and with the X-ray crystal structures. The results indicate that the solution and crystal molecular structures of cobratoxin are in substantial agreement with one another, but those of α-bungarotoxin are not. Raman data provide no evidence for strained disulfides in aqueous α-bungarotoxin, although strained CSSC dihedral angles are indicated for the X-ray crystal structure. The data are interpreted as evidence for a strained molecular conformation of α-bungarotoxin in the crystal, which converts to a relaxed, more energetically favorable conformation in aqueous solution. Raman spectra also suggest more β-strand secondary structure in aqueous α-bungarotoxin (47 ± 5%) than in the crystalline form ( < 10%). The high β-strand content measured by Raman spectroscopy could be due to either a secondary structure in solution that is appreciably different than that of the crystal, or to the imprecision of the Raman method in distinguishing peptide configurations that are vibrationally equivalent but conformationally inequivalent. Aqueous α-bungarotoxin and cobratoxin also differ from one another in amino acid side chain orientations and interactions, though not in main chain conformations. Different geometries are indicated for cystine CCSS dihedral angles, and different hydrogen bonding states are indicated for internal tyrosines. Tyrosine-24 of α-bungarotoxin is shown to donate a strong hydrogen bond to a negative acceptor, deduced to be glutamate-41, whereas the equivalently positioned residue of cobratoxin is apparently hydrogen bonded to solvent molecules.     

Truncation,modification, and optimization of MIG6segment 2 peptide to target lung cancer-related EGFR

Human epidermal growth factor receptor (EGFR) plays a central role in the pathological progression and metastasis of lung cancer; the development and clinical application of therapeutic agents that target the receptor provide important insights for new lung cancer therapies. The tumor-suppressor protein MIG6 is a negative regulator of EGFR, which can bind at the activation interface of asymmetric dimer of EGFR kinase domains to disrupt dimerization and then inactivate the kinase (Zhang X. et al. Nature 2007, 450: 741–744). The protein adopts two separated segments, i.e. MIG6^{segment 1} and MIG6^{segment 2}, to directly interact with EGFR. Here, computational modeling and analysis of the intermolecular interaction between EGFR kinase domain and MIG6^{segment 2} peptide revealed that the peptide is folded into a two-stranded β-sheet composed of β-strand 1 and β-strand 2; only the β-strand 2 can directly interact with EGFR activation loop, while leaving β-strand 1 apart from the kinase. A C-terminal island within the β-strand 2 is primarily responsible for peptide binding, which was truncated from the MIG6^{segment 2} and exhibited weak affinity to EGFR kinase domain. Structural and energetic analysis suggested that phosphorylation at residues Tyr394 and Tyr395 of truncated peptide can considerably improve EGFR affinity, and mutation of other residues can further optimize the peptide binding capability. Subsequently, three derivative versions of the truncated peptide, including phosphorylated and dephosphorylated peptides as well as a double-point mutant were synthesized and purified, and their affinities to the recombinant protein of human EGFR kinase domain were determined by fluorescence anisotropy titration. As expected theoretically, the dephosphorylated peptide has no observable binding to the kinase, and phosphorylation and mutation can confer low and moderate affinities to the peptide, respectively, suggesting a good consistence between the computational analysis and experimental assay.     

Lyotropic liquid crystals formed from ACHC-rich β-peptides

We have examined the effect of β-peptide modifications on the propensity of these helical molecules to form lyotropic liquid crystalline (LC) phases in water. All of the β-peptides we have examined contain 10 residues. In each case, at least three residues are derived from trans-2-aminocyclohexanecarboxylic acid (ACHC), which strongly promotes folding to a 14-helical conformation. The structural features varied include the number of ACHC residues, the nature and spatial arrangement of charged side chains (cationic vs anionic), and the identity of groups at the β-peptide termini. We found that relatively small changes (e.g., swapping the positions of a cationic and an anionic side chain) could have large effects, such as abrogation of LC phase formation. The trends revealed by sequence-property studies led to the design of LC-forming β-peptides that bear biomolecular recognition groups (biotin or the tripeptide Arg-Gly-Asp). Structural analysis via circular dichroism and cryo-transmission electron microscopy revealed the existence of two different types of self-associated species, globular aggregates and nanofibers. Nanofibers are the predominant assembly formed at concentrations that lead to LC phase formation, and we conclude that these nanofibers are the functional mesogens. Overall, these studies show how the modularity of β-peptide oligomers enables elucidation of the relationship between molecular structure and large-scale self-assembly behavior.     

Improved helix and kink characterization in membrane proteins allows evaluation of kink sequence predictors

Although the α-helical secondary structure of proteins is well-defined, the exact causes and structures of helical kinks are not. This is especially important for transmembrane (TM) helices of integral membrane proteins, many of which contain kinks providing functional diversity despite predominantly helical structure. We have developed a Monte Carlo method based algorithm, MC-HELAN, to determine helical axes alongside positions and angles of helical kinks. Analysis of all nonredundant high-resolution α-helical membrane protein structures (842 TM helices from 205 polypeptide chains) revealed kinks in 64% of TM helices, demonstrating that a significantly greater proportion of TM helices are kinked than those indicated by previous analyses. The residue proline is over-represented by a factor >5 if it is two or three residues C-terminal to a bend. Prolines also cause kinks with larger kink angles than other residues. However, only 33% of TM kinks are in proximity to a proline. Machine learning techniques were used to test for sequence-based predictors of kinks. Although kinks are somewhat predicted by sequence, kink formation appears to be driven predominantly by other factors. This study provides an improved view of the prevalence and architecture of kinks in helical membrane proteins and highlights the fundamental inaccuracy of the typical topological depiction of helical membrane proteins as series of ideal helices.     

Identification of drug-binding sites on human serum albumin using affinity capillary electrophoresis and chemically modified proteins as buffer additives

A technique based on affinity capillary electrophoresis (ACE) and chemically modified proteins was used to screen the binding sites of various drugs on human serum albumin (HSA). This involved using HSA as a buffer additive, following the site-selective modification of this protein at two residues (tryptophan 214 or tyrosine 411) located in its major binding regions. The migration times of four compounds (warfarin, ibuprofen, suprofen and flurbiprofen) were measured in the presence of normal or modified HSA. These times were then compared and the mobility shifts observed with the modified proteins were used to identify the binding regions of each injected solute on HSA. Items considered in optimizing this assay included the concentration of protein placed into the running buffer, the reagents used to modify HSA, and the use of dextran as a secondary additive to adjust protein mobility. The results of this method showed good agreement with those of previous reports. The advantages and disadvantages of this approach are examined, as well as its possible extension to other solutes.     

CSSP2: an improved method for predicting contact-dependent secondary structure propensity

The calculation of contact-dependent secondary structure propensity (CSSP) has been reported to sensitively detect non-native β-strand propensities in the core sequences of amyloidogenic proteins. Here we describe a noble energy-based CSSP method implemented on dual artificial neural networks that rapidly and accurately estimate the potential for the non-native secondary structure formation in local regions of protein sequences. In this method, we attempted to quantify long-range interaction patterns in diverse secondary structures by potential energy calculations and decomposition on a pairwise per-residue basis. The calculated energy parameters and seven-residue sequence information were used as inputs for artificial neural networks (ANNs) to predict sequence potential for secondary structure conversion. The trained single ANN using the >(i, i ± 4) interaction energy parameter exhibited 74% accuracy in predicting the secondary structure of test sequences in their native energy state, while the dual ANN-based predictor using (i, i ± 4) and >(i, i ± 4) interaction energies showed 83% prediction accuracy. The present method provides a simple and accurate tool for predicting sequence potential for secondary structure conversions without using 3D structural information.     

In silico characterization of antifreeze proteins using computational tools and servers

In this paper, seventeen different fish Antifreeze Proteins (AFPs) retrieved from Swiss-Prot database are analysed and characterized using in silico tools. Primary structure analysis shows that most of the AFPs are hydrophobic in nature due to the high content of non-polar residues. The presence of 11 cysteines in the rainbow smelt fish and sea raven fish AFPs infer that these proteins may form disulphide (SS) bonds, which are regarded as a positive factor for stability. The aliphatic index computed by Ex-Pasy’s ProtParam infers that AFPs may be stable for a wide range of temperature. Secondary structure analysis shows that most of the fish AFPs have predominant α-helical structures and rest of the AFPs have mixed secondary structure. The very high coil structural content of rainbow smelt fish and sea raven fish AFPs are due to the rich content of more flexible glycine and hydrophobic proline amino acids. Proline has a special property of creating kinks in polypetide chains and disrupting ordered secondary structure. SOSUI server predicts one transmembrane region in winter flounder fish and atlantic cod and two transmembrane regions in yellowtail flounder fish AFP. The predicted transmembrane regions were visualized and analysed using helical wheel plots generated by EMBOSS pepwheel tool. The presence of disulphide (SS) bonds in the AFPs Q01758 and P05140 are predicted by CYS_REC tool and also identified from the three-dimensional structure using Rasmol tool. The disulphide bonds identified from the three-dimensional structure using the Rasmol tool might be correct as the evaluation parameters are within the acceptable limits for the modelled 3D structures.     

A 1,3-phenyl-linked hydantoin oligomer scaffold as a β-strand mimetic

A synthetic scaffold that mimics a peptide β-strand has been designed and synthesised based on a 1,3-phenyl-linked hydantoin oligomer. The conformational preferences of this oligomer were investigated using molecular modelling and solution NMR experiments and suggest a planar conformation that accurately mimics the i, i + 2 and i + 4 residues of a peptide β-strand.     

Raman光谱研究铜锌超氧化物歧化酶及其金属取代衍生物的二级结构

本文测定了铜锌超氧化物歧化酶（Cu2Zn2SOD)及其金属取代衍生物Cu2Ni2SOD的Raman光谱,对图谱进行了归属,并定量测定了两种SOD的二级结构,同时对结构与活性的关系进行了讨论。     

Revealing protein structures in solid-phase peptide synthesis by 13C solid-state NMR: evidence of excessive misfolding for Alzheimer's β

Solid-phase peptide synthesis (SPPS) is a widely used technique in biology and chemistry. However, the synthesis yield in SPPS often drops drastically for longer amino acid sequences, presumably because of the occurrence of incomplete coupling reactions. The underlying cause for this problem is hypothesized to be a sequence-dependent propensity to form secondary structures through protein aggregation. However, few methods are available to study the site-specific structure of proteins or long peptides that are anchored to the solid support used in SPPS. This study presents a novel solid-state NMR (SSNMR) approach to examine protein structure in the course of SPPS. As a useful benchmark, we describe the site-specific SSNMR structural characterization of the 40-residue Alzheimer's β-amyloid (Aβ) peptide during SPPS. Our 2D (13)C/(13)C correlation SSNMR data on Aβ(1-40) bound to a resin support demonstrated that Aβ underwent excessive misfolding into a highly ordered β-strand structure across the entire amino acid sequence during SPPS. This approach is likely to be applicable to a wide range of peptides/proteins bound to the solid support that are synthesized through SPPS.     

An in silico analysis of primary and secondary structure specificity determinants for human peptidylarginine deiminase types 2 and 4

Human peptidylarginine deiminases (hPADs) are a family of five calcium-dependent enzymes that facilitate citrullination, which is the post-translational modification of peptidyl arginine to peptidyl citrulline. The isozymes hPAD2 and hPAD4 have been implicated in the development and progression of several autoimmune diseases, including rheumatoid arthritis and multiple sclerosis. To better characterize the primary and secondary structure determinants of citrullination specificity, we mined the literature for protein sequences susceptible to citrullination by hPAD2 or hPAD4. First, protein secondary structure classification (α-helix, β-sheet, or coil) was predicted using the PSIPRED software. Next, we used motif-x and pLogo to extract and visualize statistically significant motifs within each data set. Within the data sets of peptides predicted to lie in coil regions, both hPAD2 and hPAD4 appear to favor citrullination of glycine-containing motifs, while distinct hydrophobic motifs were identified for hPAD2 citrullination sites predicted to reside within α-helical and β-sheet regions. Additionally, we identified potential substrate overlap between coil region citrullination and arginine methylation. Together, these results confirm the importance and offer some insight into the role of secondary structure elements for citrullination specificity, and provide biological context for the existing hPAD specificity and arginine post-translational modification literature.     

Structure and Composition of Insulin Fibril Surfaces Probed by TERS

Amyloid fibrils associated with many neurodegenerative diseases are the most intriguing targets of modern structural biology. Significant knowledge has been accumulated about the morphology and fibril-core structure recently. However, no conventional methods could probe the fibril surface despite its significant role in the biological activity. Tip-enhanced Raman spectroscopy (TERS) offers a unique opportunity to characterize the surface structure of an individual fibril due to a high depth and lateral spatial resolution of the method in the nanometer range. Herein, TERS is utilized for characterizing the secondary structure and amino acid residue composition of the surface of insulin fibrils. It was found that the surface is strongly heterogeneous and consists of clusters with various protein conformations. More than 30% of the fibril surface is dominated by β-sheet secondary structure, further developing Dobson's model of amyloid fibrils (Jimenez et al. Proc. Natl. Acad. Sci. U.S.A. 2002 , 99 , 9196 - 9201 ). The propensity of various amino acids to be on the fibril surface and specific surface secondary structure elements were evaluated. β-sheet areas are rich in cysteine and aromatic amino acids, such as phenylalanine and tyrosine, whereas proline was found only in α-helical and unordered protein clusters. In addition, we showed that carboxyl, amino, and imino groups are nearly equally distributed over β-sheet and α-helix/unordered regions. Overall, this study provides valuable new information about the structure and composition of the insulin fibril surface and demonstrates the power of TERS for fibril characterization.