Search

Deep learning is a class of machine learning techniques capable of creating internal representation of data without explicit preprogramming. Hence, in addition to practical applications, it is of interest to analyze what features of biological data may be learned by such models. Here, we describe PredPair, a deep learning neural network trained to predict base pairs in RNA structure from sequence alone, without any incorporated prior knowledge, such as the stacking energies or possible spatial structures. PredPair learned the Watson-Crick and wobble base-pairing rules and created an internal representation of the stacking energies and helices. Application to independent experimental (DMS-Seq) data on nucleotide accessibility in mRNA showed that the nucleotides predicted as paired indeed tend to be involved in the RNA structure. The performance of the constructed model was comparable with the state-of-the-art method based on the thermodynamic approach, but with a higher false positives rate. On the other hand, it successfully predicted pseudoknots. t-SNE clusters of embeddings of RNA sequences created by PredPair tend to contain embeddings from particular Rfam families, supporting the predictions of PredPair being in line with biological classification., Competing Interests: Mikhail Gelfand is an Academic Editor for PeerJ., (© 2022 Grigorashvili et al.)

As has been recently shown, the toxicity of protein aggregates is determined by their structure. Therefore, special attention has been focused on the search for factors that specify the structural features of formed amyloid fibrils. The effect of amino acid substitutions in apomyoglobin on the structural characteristics of its amyloid aggregates has been analyzed. The morphology and secondary structure of amyloids of the wild-type protein and its mutant variants Val10Ala, Val10Phe, and Trp14Phe have been compared, and the regions involved in intermolecular interactions in fibrils have been determined using limited proteolysis and mass spectrometry. No considerable differences have been found in the morphology (shape, length, or diameter) or the content (percentage) of the cross-β structure of apomyoglobin amyloids and its mutant variants. Amyloid cores of wild-type apomyoglobin and variants with Val10Phe and Trp14Phe substitutions have been formed by different regions of the polypeptide chain. The case study of apomyoglobin demonstrates that the location of amyloidogenic regions in the polypeptide chain of wild-type protein and its mutant forms can differ. Thus, possible structural changes in amyloids resulting from amino acid substitutions should be taken into account when studying phenotype aggregation.

We performed a comparative study of the process of amyloid formation by short homologous peptides with a substitution of aspartate for glutamate in position 2 - VDSWNVLVAG (AspNB) and VESWNVLVAG (GluNB) - with unblocked termini. Peptide AspNB (residues 166-175) corresponded to the predicted amyloidogenic region of the protein glucantransferase Bgl2 from the Saccharomyces cerevisiae cell wall. The process of amyloid formation was monitored by fluorescence spectroscopy (FS), electron microscopy (EM), tandem mass spectrometry (TMS), and X-ray diffraction (XD) methods. The experimental study at pH3.0 revealed formation of amyloid fibrils with similar morphology for both peptides. Moreover, we found that the morphology of fibrils made of untreated ammonia peptide is not mentioned in the literature. This morphology resembles snakes lying side by side in the form of a wave without intertwining. Irrespective of the way of the peptide preparation, the rate of fibril formation is higher for AspNB than for GluNB. However, preliminary treatment with ammonia highly affected fibril morphology especially for AspNB. Such treatment allowed us to obtain a lag period during the process of amyloid formation. It showed that the process was nucleation-dependent. With or without treatment, amyloid fibrils consisted of ring-like oligomers with the diameter of about 6nm packed either directly ring-to-ring or ring-on-ring with a slight shift. We also proposed the molecular structure of amyloid fibrils for two studied peptides., (Copyright © 2016 Elsevier B.V. All rights reserved.)

It has been demonstrated using Aβ40 and Aβ42 recombinant and synthetic peptides that their fibrils are formed of complete oligomer ring structures. Such ring structures have a diameter of about 8-9 nm, an oligomer height of about 2- 4 nm, and an internal diameter of the ring of about 3-4 nm. Oligomers associate in a fibril in such a way that they interact with each other, overlapping slightly. There are differences in the packing of oligomers in fibrils of recombinant and synthetic Aβ peptides. The principal difference is in the degree of orderliness of ring-like oligomers that leads to generation of morphologically different fibrils. Most ordered association of ring-like structured oligomers is observed for a recombinant Aβ40 peptide. Less ordered fibrils are observed with the synthetic Aβ42 peptide. Fragments of fibrils the most protected from the action of proteases have been determined by tandem mass spectrometry. It was shown that unlike Aβ40, fibrils of Aβ42 are more protected, showing less ordered organization compared to that of Aβ40 fibrils. Thus, the mass spectrometry data agree with the electron microscopy data and structural models presented here.

The studies of amyloid structures and the process of their formation are important problems of biophysics. One of the aspects of such studies is to determine the amyloidogenic regions of a protein chain that form the core of an amyloid fibril. We have theoretically predicted the amyloidogenic regions of the Aβ(1-40) peptide capable of forming an amyloid structure. These regions are from 16 to 21 and from 32 to 36 amino acid residues. In this work, we have attempted to identify these sites experimentally by the method of tandem mass spectrometry. As a result, we show that regions of the Aβ(1-40) peptide from 16 to 22 and from 28 to 40 amino acid residues are resistant to proteases, i.e. they are included in the core of amyloid fibrils. Our results correlate with the results of the theoretical prediction.

A method for the synthesis and high purification of fragments of Aβ(1-42) peptide has been elaborated. We have synthesized the amyloidogenic fragment Aβ(16-25) predicted by us and studied the process of its aggregation by electron microscopy and X-ray analysis. Electron microscopy images show that the peptide forms a film, which is not characteristic of amyloid fibrils. At the same time, according to the X-ray diffraction data, its preparations display the presence of two main reflections (4.6-4.8 and 8-12 Å) characteristic of cross-β structure of amyloid fibrils. Thus, the fragment Aβ(16-25) that we predicted is a promising object not only for studying the process of polymerization of the peptides/proteins, but also for using it as a nanomaterial to study a number of biological processes.

In the presented paper, theoretical as well as electron microscopy and X-ray diffraction experimental approaches were employed for studding the process of Aβ amyloid formation. Using quantitative estimates of a number of monomers which form the nuclei of amyloid fibrils the sizes of folding nuclei of amyloid fibrils for Aβ40 and 42 have been determined for the first time. We have shown that the size of the primary nucleus of Aβ42 peptide fibrils corresponds to 3 monomers, the size of the secondary nucleus for this peptide is 2 monomers. Applying the same analysis to Aβ40 we conclude that the size of the primary nucleus is 2 monomers, and the size of the secondary nucleus is one monomer. Summation of our theoretical and experimental results has allowed us to propose a new model of the structural organization of amyloid fibrils. Our model suggests that the generation of fibrils takes place along the following simplified pathway: a monomer→a ring oligomer→a mature fibril consisting of ring oligomers. These data shed more light upon our understanding of what sizes of the oligomers could represent main targets for future therapies (tetramers for Aβ42 and trimers for Aβ40), and aid in the development of inhibitors of Aβ40 and 42 oligomer formation., (Copyright © 2016. Published by Elsevier Inc.)

The data presented in this article are related to the research article entitled "One of the possible mechanisms of amyloid fibrils formation based on the sizes of primary and secondary folding nuclei of Aβ40 and Aβ42" (Dovidchenko et al., 2016) [1]. Aβ peptide is one of the most intensively studied amyloidogenic peptides. Despite the huge number of articles devoted to studying different fragments of Aβ peptide there are only several papers with correct kinetics data, also there are a few papers with X-ray data, especially for Aβ42. Our data present X-ray diffraction patterns both for Aβ40 and Aβ42 as well for Tris-HCl and wax. Moreover, our data provide kinetics of amyloid formation by recombinant Аβ40 and synthetic Аβ42 peptides by using electron microscopy.

We have developed a highly efficient method for purification of the recombinant product Aβ(1-40) peptide. The concentration dependence of amyloid formation by recombinant Aβ(1-40) peptide was studied using fluorescence spectroscopy and electron microscopy. We found that the process of amyloid formation is preceded by lag time, which indicates that the process is nucleation-dependent. Further exponential growth of amyloid fibrils is followed by branching scenarios. Based on the experimental data on the concentration dependence, the sizes of the folding nuclei of fibrils were calculated. It turned out that the size of the primary nucleus is one "monomer" and the size of the secondary nucleus is zero. This means that the nucleus for new aggregates can be a surface of the fibrils themselves. Using electron microscopy, we have demonstrated that fibrils of these peptides are formed by the association of rounded ring structures.

The aim of this study was to investigate the process of amyloidogenesis of amyloid-β (Aβ)42 peptide, by means of fluorescence spectroscopy, electron microscopy, X-ray diffraction, and mass spectrometry. It has been repeatedly reported in the literature that the process of fibril formation by Aβ42 peptide depends considerably not only upon the specific conditions (ionic conditions, pH, temperature, mixing, etc.), as well as the manufacturing route (synthetic or recombinant), but also on the methods of synthesis and purification. We have, for the first time, systematically analyzed samples of Aβ42 peptide supplied by five different companies (Anaspec, Invitrogen, Enzo, Sigma-Aldrich, and SynthAssist) and obtained evidence of significant variability, including lot to lot variations. All studied samples formed amyloid-like fibrils at pH3-6, and the fibrils contained cross-β structures. Samples from Anaspec, Invitrogen, and Enzo formed one particular type of amyloid-like fibrils, while the samples from Sigma-Aldrich and SynthAssist formed another distinct type of fibrils. The observed polymorphism emphasizes the capacity of the Aβ42 peptide to act as a prion agent with varying structural characteristics. The presented data have allowed us to propose a possible mechanism of formation of amyloid-like fibrils.