Structure Prediction Of Peptides Research Articles

Advances in Peptide/Protein Structure Prediction Tools and their Relevance for Structural Biology in the Last Decade

Abstract: Peptides and proteins are involved in several biological processes at a molecular level. In this context, three-dimensional structure characterization and determination of peptides and proteins have helped researchers unravel the chemical and biological role of these macromolecules. Over 50 years, peptide and protein structures have been determined by experimental methods, including nuclear magnetic resonance (NMR), X-ray crystallography, and cryo-electron microscopy (cryo-EM). Therefore, an increasing number of atomic coordinates for peptides and proteins have been deposited in public databases, thus assisting the development of computational tools for predicting unknown 3D structures. In the last decade, a race for innovative methods has arisen in computational sciences, including more complex biological activity and structure prediction algorithms. As a result, peptide/protein theoretical models have achieved a new level of structure prediction accuracy compared with experimentally determined structures. Machine learning and deep learning approaches, for instance, incorporate fundamental aspects of peptide/protein geometry and include physical/biological knowledge about these macromolecules' experimental structures to build more precise computational models. Additionally, computational strategies have helped structural biology, including comparative, threading, and ab initio modeling and, more recently, prediction tools based on machine learning and deep learning. Bearing this in mind, here we provide a retrospective of protein and peptide structure prediction tools, highlighting their advances and obstacles and how they have assisted researchers in answering crucial biological questions.

PEP-FOLD4: a pH-dependent force field for peptide structure prediction in aqueous solution.

Accurate and fast structure prediction of peptides of less 40 amino acids in aqueous solution has many biological applications, but their conformations are pH- and salt concentration-dependent. In this work, we present PEP-FOLD4 which goes one step beyond many machine-learning approaches, such as AlphaFold2, TrRosetta and RaptorX. Adding the Debye-Hueckel formalism for charged-charged side chain interactions to a Mie formalism for all intramolecular (backbone and side chain) interactions, PEP-FOLD4, based on a coarse-grained representation of the peptides, performs as well as machine-learning methods on well-structured peptides, but displays significant improvements for poly-charged peptides. PEP-FOLD4 is available at http://bioserv.rpbs.univ-paris-diderot.fr/services/PEP-FOLD4. This server is free and there is no login requirement.

A refined pH-dependent coarse-grained model for peptide structure prediction in aqueous solution.

Introduction: Peptides carry out diverse biological functions and the knowledge of the conformational ensemble of polypeptides in various experimental conditions is important for biological applications. All fast dedicated softwares perform well in aqueous solution at neutral pH. Methods: In this study, we go one step beyond by combining the Debye-Hückel formalism for charged-charged amino acid interactions and a coarse-grained potential of the amino acids to treat pH and salt variations. Results: Using the PEP-FOLD framework, we show that our approach performs as well as the machine-leaning AlphaFold2 and TrRosetta methods for 15 well-structured sequences, but shows significant improvement in structure prediction of six poly-charged amino acids and two sequences that have no homologous in the Protein Data Bank, expanding the range of possibilities for the understanding of peptide biological roles and the design of candidate therapeutic peptides.

Benchmarking AlphaFold2 on peptide structure prediction

Recent advancements in computational tools have allowed protein structure prediction with high accuracy. Computational prediction methods have been used for modeling many soluble and membrane proteins, but the performance of these methods in modeling peptide structures has not yet been systematically investigated. We benchmarked the accuracy of AlphaFold2 in predicting 588 peptide structures between 10 and 40 amino acids using experimentally determined NMR structures as reference. Our results showed AlphaFold2 predicts α-helical, β-hairpin, and disulfide-rich peptides with high accuracy. AlphaFold2 performed at least as well if not better than alternative methods developed specifically for peptide structure prediction. AlphaFold2 showed several shortcomings in predicting Φ/Ψ angles, disulfide bond patterns, and the lowest RMSD structures failed to correlate with lowest pLDDT ranked structures. In summary, computation can be a powerful tool to predict peptide structures, but additional steps may be necessary to analyze and validate the results.

A Generalized Attraction-Repulsion Potential and Revisited Fragment Library Improves PEP-FOLD Peptide Structure Prediction.

Fast and accurate structure prediction is essential to the study of peptide function, molecular targets, and interactions and has been the subject of considerable efforts in the past decade. In this work, we present improvements to the popular simplified PEP-FOLD technique for small peptide structure prediction. PEP-FOLD originality is threefold: (i) it uses a predetermined structural alphabet, (ii) it uses a sequential algorithm to reconstruct the tridimensional structures of these peptides in a discrete space using a fragment library, and (iii) it assesses the energy of these structures using a coarse-grained representation in which all of the backbone atoms but the α-hydrogen are present, and the side chain corresponds to a unique bead. In former versions of PEP-FOLD, a van der Waals formulation was used for non-bonded interactions, with each side chain being associated with a fixed radius. Here, we explore the relevance of using instead a generalized formulation in which not only the optimal distance of interaction and the energy at this distance are parameters but also the distance at which the potential is zero. This allows each side chain to be associated with a different radius and potential energy shape, depending on its interaction partner, and in principle to make more effective the coarse-grained representation. In addition, the new PEP-FOLD version is associated with an updated library of fragments. We show that these modifications lead to important improvements for many of the problematic targets identified with the former PEP-FOLD version while maintaining already correct predictions. The improvement is in terms of both model ranking and model accuracy. We also compare the PEP-FOLD enhanced version to state-of-the-art techniques for both peptide and structure predictions: APPTest, RaptorX, and AlphaFold2. We find that the new predictions are superior, in particular with respect to the prediction of small β-targets, to those of APPTest and RaptorX and bring, with its original approach, additional understanding on folded structures, even when less precise than AlphaFold2. With their strong physical influence, the revised structural library and coarse-grained potential offer, however, the means for a deeper understanding of the nature of folding and open a solid basis for studying flexibility and other dynamical properties not accessible to IA structure prediction approaches.

An Integrated In Silico Approach to Develop Epitope-Based Peptide Vaccine Against SARS-CoV-2

Background: SARS-CoV-2 has been the talk of the town ever since the beginning of 2020. Every country is trying all possible steps to combat the disease ranging from shutting the complete economy of the country to the repurposing of drugs and vaccine development. The rapid data analysis and widespread tools have made bioinformatics capable of giving new insights to deal with the current scenario more efficiently through an emerging field, Vaccinomics. Objective: The present in-silico study was attempted to identify peptide fragments from spike surface glycoprotein of SARS-CoV-2 that can be efficiently used for the development of an epitope-based vaccine designing approach. Methodology: The epitopes of B and T-cell are predicted using integrated computational tools. VaxiJen server, NetCTL, and IEDB tools were used to study, analyze, and predict potent T-cell epitopes, its subsequent MHC-I interactions, and B-cell epitopes. The 3D structure prediction of peptides and MHC-I alleles (HLA-C*03:03) was further done using AutoDock4.0. Result: Based on result interpretation, the peptide sequence from 1138-1145 amino acid and sequence WTAGAAAYY and YDPLQPEL were obtained as potential B-cell and T-cell epitopes respectively. Conclusion: The peptide sequence WTAGAAAYY and the amino acid sequence from 1138-1145 of the spike protein of SARS-CoV-2 can be used as a probable B-cell epitope candidate. Also, the amino acid sequence YDPLQPEL can be used as a potent T-cell epitope. This in-silico study will help us to identify novel epitope-based peptide vaccine targets in the spike protein of SARS-CoV-2. Further, the in-vitro and in-vivo study needed to validate the findings.

Accelerated cyclic peptide ensemble generation using reservoir REMD

Cyclic peptides have been widely used as potential therapeutics and scaffolded antigens due to their ability to bias the conformation to the desired structure using just a few amino acids. Although many tools have been developed to perform cyclic peptide structure prediction, there is little discussion on the ensemble generation of the cyclic peptides. We found that conventional molecular dynamics (MD) simulation is insufficient to sample the ensemble of small cyclic peptides (<10 amino acids), likely due to energy barriers rooted in the ring-like conformational constraint.

APPTEST is a novel protocol for the automatic prediction of peptide tertiary structures.

Good knowledge of a peptide’s tertiary structure is important for understanding its function and its interactions with its biological targets. APPTEST is a novel computational protocol that employs a neural network architecture and simulated annealing methods for the prediction of peptide tertiary structure from the primary sequence. APPTEST works for both linear and cyclic peptides of 5–40 natural amino acids. APPTEST is computationally efficient, returning predicted structures within a number of minutes. APPTEST performance was evaluated on a set of 356 test peptides; the best structure predicted for each peptide deviated by an average of 1.9Å from its experimentally determined backbone conformation, and a native or near-native structure was predicted for 97% of the target sequences. A comparison of APPTEST performance with PEP-FOLD, PEPstrMOD and PepLook across benchmark datasets of short, long and cyclic peptides shows that on average APPTEST produces structures more native than the existing methods in all three categories. This innovative, cutting-edge peptide structure prediction method is available as an online web server at https://research.timmons.eu/apptest, facilitating in silico study and design of peptides by the wider research community.

Advancing Rational Control of Peptide-Surface Complexes.

Understanding peptide-surface interactions is crucial for programming self-assembly of peptides at surfaces and in realizing their applications, such as biosensors and biomimetic materials. In this study, we developed insights into the dependence of a residue's interaction with a surface on its neighboring residue in a tripeptide using molecular dynamics simulations. This knowledge is integral for designing rational mutations to control peptide-surface complexes. Using graphene as our model surface, we estimated the free energy of adsorption (ΔAads) and extracted predominant conformations of 26 tripeptides with the motif LNR-CR-Gly, where LNR and CR are variable left-neighboring and central residues, respectively. We considered a combination of strongly adsorbing (Phe, Trp, and Arg) and weakly adsorbing (Ala, Val, Leu, Ser, and Thr) amino acids on graphene identified in a prior study to form the tripeptides. Our results indicate that ΔAads of a tripeptide cannot be estimated as the sum of ΔAads of each residue indicating that the residues in a tripeptide do not behave as independent entities. We observed that the contributions from the strongly adsorbing amino acids were dominant, which suggests that such residues could be used for strengthening peptide-graphene interactions irrespective of their neighboring residues. In contrast, the adsorption of weakly adsorbing central residues is dependent on their neighboring residues. Our structural analysis revealed that the dihedral angles of LNR are more correlated with that of CR in the adsorbed state than in bulk state. Together with ΔAads trends, this implies that different backbone structures of a given CR can be accessed for a similar ΔAads by varying the LNR. Therefore, incorporation of context effects in designing mutations can lead to desired peptide structure at surfaces. Our results also emphasize that these cooperative effects in ΔAads and structure are not easily predicted a priori. The collective results have applications in guiding rational mutagenesis techniques to control orientation of peptides at surfaces and in developing peptide structure prediction algorithms in adsorbed state from its sequence.

CRiSP: accurate structure prediction of disulfide-rich peptides with cystine-specific sequence alignment and machine learning.

High-throughput sequencing discovers many naturally occurring disulfide-rich peptides or cystine-rich peptides (CRPs) with diversified bioactivities. However, their structure information, which is very important to peptide drug discovery, is still very limited. We have developed a CRP-specific structure prediction method called Cystine-Rich peptide Structure Prediction (CRiSP), based on a customized template database with cystine-specific sequence alignment and three machine-learning predictors. The modeling accuracy is significantly better than several popular general-purpose structure modeling methods, and our CRiSP can provide useful model quality estimations. The CRiSP server is freely available on the website at http://wulab.com.cn/CRISP. wuyd@pkusz.edu.cn or jiangfan@pku.edu.cn. Supplementary data are available at Bioinformatics online.

Evaluating the peptide structure prediction capabilities of a purely ab-initio method.

DEEPSAM is a relatively new global optimization algorithm aimed to predict the structure of bio-molecules from sequence, without any additional preliminary assumption. It is an evolutionary algorithm whose mutation operators are built by hybridizing the diffusion equation method, molecular dynamics simulated annealing, and a quasi-Newton local minimization method. The goal of this study was to evaluate the structure prediction capabilities of DEEPSAM by running it upon NMR structures of linear peptides (10-20 residues). The results indicate that DEEPSAM successfully predicted the conformations of these peptides, using modest computing resources.

PEPstrMOD: structure prediction of peptides containing natural, non-natural and modified residues.

BackgroundIn the past, many methods have been developed for peptide tertiary structure prediction but they are limited to peptides having natural amino acids. This study describes a method PEPstrMOD, which is an updated version of PEPstr, developed specifically for predicting the structure of peptides containing natural and non-natural/modified residues.ResultsPEPstrMOD integrates Forcefield_NCAA and Forcefield_PTM force field libraries to handle 147 non-natural residues and 32 types of post-translational modifications respectively by performing molecular dynamics using AMBER. AMBER was also used to handle other modifications like peptide cyclization, use of D-amino acids and capping of terminal residues. In addition, GROMACS was used to implement 210 non-natural side-chains in peptides using SwissSideChain force field library. We evaluated the performance of PEPstrMOD on three datasets generated from Protein Data Bank; i) ModPep dataset contains 501 non-natural peptides, ii) ModPep16, a subset of ModPep, and iii) CyclicPep contains 34 cyclic peptides. We achieved backbone Root Mean Square Deviation between the actual and predicted structure of peptides in the range of 3.81–4.05 Å.ConclusionsIn summary, the method PEPstrMOD has been developed that predicts the structure of modified peptide from the sequence/structure given as input. We validated the PEPstrMOD application using a dataset of peptides having non-natural/modified residues. PEPstrMOD offers unique advantages that allow the users to predict the structures of peptides having i) natural residues, ii) non-naturally modified residues, iii) terminal modifications, iv) post-translational modifications, v) D-amino acids, and also allows extended simulation of predicted peptides. This will help the researchers to have prior structural information of modified peptides to further design the peptides for desired therapeutic property. PEPstrMOD is freely available at http://osddlinux.osdd.net/raghava/pepstrmod/.ReviewersThis article was reviewed by Prof Michael Gromiha, Dr. Bojan Zagrovic and Dr. Zoltan Gaspari.Electronic supplementary materialThe online version of this article (doi:10.1186/s13062-015-0103-4) contains supplementary material, which is available to authorized users.

Early oligomerization stages for the non-amyloid component of α-synuclein amyloid.

In recent years, much effort has focused on the early stages of aggregation and the formation of amyloid oligomers. Aggregation processes for these proteins are complex and their non-equilibrium nature makes any experimental study very difficult. Under these conditions, simulations provide a useful alternative for understanding the dynamics of the early stages of oligomerization. Here, we focus on the non-Aβ amyloid component (NAC) of the monomer, dimer, and trimer of α-synuclein, an important 35-residue sequence involved in the aggregation and fibrillation of this protein associated with Parkinson's disease. Using Hamiltonian and temperature replica exchange molecular dynamics simulations combined with the coarse grained Optimized Potential for Efficient peptide structure Prediction potential, we identify the role of the various regions and the secondary structures for the onset of oligomerization. For this sequence, we clearly observe the passage from α-helix to β-sheet, a characteristic transition of amyloid proteins. More precisely, we find that the NAC monomer is highly structured with two α-helical regions, between residues 2-13 and 19-25. As the dimer and trimer form, β-sheet structures between residues 2-14 and 26-34 appear and rapidly structure the system. The resulting conformations are much more structured than similar dimers and trimers of β-amyloid and amylin proteins and yet display a strong polymorphism at these early stages of aggregation. In addition to its inherent experimental interest, comparison with other sequences shows that NAC could be a very useful numerical model for understanding the onset of aggregation.

Exploring amyloid aggregates with coarse-grained protein simulations

ABSTRACTProteins are complex, yet elegant, machines fine-tuned by evolution to properly fulfill a variety of tasks in the crowded cellular environment. These are, however, very challenging numerically due to their dimension, number of degrees of freedom and the wide range of relevant time scales. With aging, some proteins misfold and form harmful amyloid aggregates associated with multiple neurodegenerative diseases, and in particular Alzheimer’s, which challenge our society today. Here, I present the coarse-grained OPEP (Optimized Potential for Efficient peptide structure Prediction) force field and what we can learn from OPEP simulations to get insights into the self-assembly of amyloid peptides.

PEP-FOLD: an updated de novo structure prediction server for both linear and disulfide bonded cyclic peptides

In the context of the renewed interest of peptides as therapeutics, it is important to have an on-line resource for 3D structure prediction of peptides with well-defined structures in aqueous solution. We present an updated version of PEP-FOLD allowing the treatment of both linear and disulphide bonded cyclic peptides with 9–36 amino acids. The server makes possible to define disulphide bonds and any residue–residue proximity under the guidance of the biologists. Using a benchmark of 34 cyclic peptides with one, two and three disulphide bonds, the best PEP-FOLD models deviate by an average RMS of 2.75 Å from the full NMR structures. Using a benchmark of 37 linear peptides, PEP-FOLD locates lowest-energy conformations deviating by 3 Å RMS from the NMR rigid cores. The evolution of PEP-FOLD comes as a new on-line service to supersede the previous server. The server is available at: http://bioserv.rpbs.univ-paris-diderot.fr/PEP-FOLD.

Peptide structure prediction using distributed volunteer computing networks

Recent investigations to develop novel antimicrobial, antibiotical drugs have focused on the development of artificial protein peptides. As short peptides are naturally involved in many important biological processes in the cell and therefore target many kinds of cells. To functionalize peptides it is vital to design peptides, which can differentially target bacterial and eucariotic cells. Although the length of the peptides investigated in this study was limited to 16 amino acids, the number of possible peptide sequences is still too large to synthesize them in a trial- and error manner, therefore requiring a method for directed, but also high-througput peptide design. By predicting the structure of peptide proteins, this design process can be supported through structure-function analysis and peptide-membrane interaction simulation. In this investigation we could predict peptide structures de-novo, i.e. with the sequence information alone, using a massively parallel simulation scheme. We sample a sizable fraction of the peptide’s conformational space using Monte-Carlo simulations in the free-energy forcefield PFF02 on the volunteer computing network POEM@HOME. This forcefield models the protein’s native conformation as the global minimum of the free-energy. We could identify peptides of different topologies in a completely automated manner, which allows for the high-throughput screening of large peptide databases for their structural features, which would allow the rapid protopying of peptides needed for novel peptide design.

Structure and Thermodynamics of Amylin Dimer Studied by Hamiltonian-Temperature Replica Exchange Molecular Dynamics Simulations

The loss of the insulin-producing β-cells in the pancreatic islets of Langerhans, responsible for type-II diabetes, is associated with islet amyloid deposits. The main component of these deposits is the amyloid fibrils formed by the 37-residue human islet amyloid polypeptide (hIAPP also known as amylin). Although the fibrils are well characterized by cross β structure, the structure of the transient oligomers formed in the early stage of aggregation remains elusive. In this study, we apply the Hamiltonian-temperature replica exchange molecular dynamics to characterize the structure and thermodynamics of a full-length hIAPP dimer in both the presence and the absence of the Cys2-Cys7 disulfide bond. We compare these results with those obtained on the monomeric and dimeric forms of rat IAPP (rIAPP) with a disulfide bridge which differ from the hIAPP by 6 amino acids in the C-terminal region, but it is unable to form fibrils. Using a coarse-grained protein force field (OPEP-the Optimized Potential for Efficient peptide structure Prediction) running for a total of 10-28 μs per system studied, we show that sequences sample α-helical structure in the N-terminal region but that the length of this secondary element is shorter and less stable for the chains without the disulfide bridge (residues 5-16 for hIAPP with the bridge vs 10-16 for hIAPP without the bridge). This α-helix is known to be an important transient stage in the formation of oligomers. In the C-terminal, the amyloidogenic region of hIAPP, β-strands are seen for residues 17-26 and 30-35. On the contrary, no significant β-sheet content in the C-terminal is observed for either the monomeric or the dimeric rIAPP. These numerical results are fully consistent with recent experimental findings that the N-terminal residues are not part of the fibril by forming α-helical structure but rather play a significant role in stabilizing the amyloidogenic region available for the fibrillation.

Cancer Osaka thyroid (Cot) phosphorylates Polo-like kinase (PLK1) at Ser137 but not at Thr210

Cancer Osaka thyroid (Cot) is a proto-oncogenic kinase which belongs to the MAP3K family. A peptide-based substrate screening assay revealed that Cot has the ability to phosphorylate Polo-like kinase 1 (Plk1) at Ser137. Kinase assays with intact Plk1 and peptides surrounding Ser137 and Thr210 indicated further that Cot phosphorylates Ser137 but not Thr210. Additional support came from 3D peptide structure prediction and Cot-Plk1 interaction modeling. In vivo experiments demonstrated that wild type Cot, but not a kinase-dead mutant, has the ability to phosphorylate Ser137. Knockdown of Cot in Hela showed a reduction in the level of phosphorylation of Ser137. These results imply for the first time that Cot might be an upstream kinase of Plk1 and suggest a new mechanism for the regulation of the cellular function of Plk1.

Generalized Pattern Search Algorithm for Peptide Structure Prediction

Finding the near-native structure of a protein is one of the most important open problems in structural biology and biological physics. The problem becomes dramatically more difficult when a given protein has no regular secondary structure or it does not show a fold similar to structures already known. This situation occurs frequently when we need to predict the tertiary structure of small molecules, called peptides. In this research work, we propose a new ab initio algorithm, the generalized pattern search algorithm, based on the well-known class of Search-and-Poll algorithms. We performed an extensive set of simulations over a well-known set of 44 peptides to investigate the robustness and reliability of the proposed algorithm, and we compared the peptide conformation with a state-of-the-art algorithm for peptide structure prediction known as PEPstr. In particular, we tested the algorithm on the instances proposed by the originators of PEPstr, to validate the proposed algorithm; the experimental results confirm that the generalized pattern search algorithm outperforms PEPstr by 21.17% in terms of average root mean-square deviation, RMSD C α .

The complex folding pathways of protein A suggest a multiple-funnelled energy landscape

Folding proteins into their native states requires the formation of both secondary and tertiary structures. Many questions remain, however, as to whether these form into a precise order, and various pictures have been proposed that place the emphasis on the first or the second level of structure in describing folding. One of the favorite test models for studying this question is the B domain of protein A, which has been characterized by numerous experiments and simulations. Using the activation-relaxation technique coupled with a generic energy model (optimized potential for efficient peptide structure prediction), we generate more than 50 folding trajectories for this 60-residue protein. While the folding pathways to the native state are fully consistent with the funnel-like description of the free energy landscape, we find a wide range of mechanisms in which secondary and tertiary structures form in various orders. Our nonbiased simulations also reveal the presence of a significant number of non-native beta and alpha conformations both on and off pathway, including the visit, for a non-negligible fraction of trajectories, of fully ordered structures resembling the native state of nonhomologous proteins.