Share This Paper. Background Citations. Methods Citations. Results Citations. Figures, Tables, and Topics from this paper. Citation Type. Has PDF. Publication Type. More Filters. Highly Influenced. View 5 excerpts, cites results. An aggregate has a very regular beta structure with a tightly packed core, which … Expand.
Sequence- and structure-based prediction of amyloidogenic regions in proteins. Amyloid fibrils from muscle myoglobin. Fantini, J. Structural and molecular modelling studies reveal a new mechanism of action of chloroquine and hydroxychloroquine against SARS-CoV-2 infection. Agents Gazit, E.
A possible role for pi-stacking in the self-assembly of amyloid fibrils. Self assembly of short aromatic peptides into amyloid fibrils and related nanostructures. Prion 1, 32— Hannestad, J. Hasanbasic, S. Prolines affect the nucleation phase of amyloid fibrillation reaction; Mutational analysis of human stefin B. ACS Chem. Hashimoto, Y. Effects of glycosylation on the structure and dynamics of eel calcitonin in micelles and lipid bilayers determined by nuclear magnetic resonance spectroscopy.
Biochemistry 38, — Hayat, S. Bioinformatics 32, — Hofmann, K. TMbase - a database of membrane spanning proteins segments. Hoppe Seyler Google Scholar. Jacoboni, I. Prediction of the transmembrane regions of beta barrel membrane proteins with a neural network based predictor. Jamasbi, E. Fluorescence imaging of the interaction of amyloid beta 40 peptides with live cells and model membrane.
Acta Biomembr. Jang, H. Computational methods for structural and functional studies of Alzheimer's amyloid ion channels. Methods Mol. Jenko, S. Different propensity to form amyloid fibrils by two homologous proteins-Human stefins A and B: searching for an explanation.
Proteins 55, — Julien, C. Acta Neuropatholog. Kahlenberg, J. Little peptide, big effects: the role of LL in inflammation and autoimmune disease. Kametani, F. Reconsideration of amyloid hypothesis and tau hypothesis in Alzheimer's disease.
Kandel, N. B , — Kawahara, M. Alzheimer's beta-amyloid, human islet amylin, and prion protein fragment evoke intracellular free calcium elevations by a common mechanism in a hypothalamic GnRH neuronal cell line. Alzheimer's Dis. Kayed, R. Soluble endogenous oligomeric alpha-synuclein species in neurodegenerative diseases: expression, spreading, and cross-talk.
Parkinsons Dis. Kreutzer, A. Krogh, A. Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. Lashuel, H. Neurodegenerative disease: amyloid pores from pathogenic mutations affiliations. Nature Are amyloid diseases caused by protein aggregates that mimic bacterial pore-forming toxins?
Last, N. Common mechanism unites membrane poration by amyloid and antimicrobial peptides. Lee, J. Amyloid beta ion channels in a membrane comprising brain total lipid extracts. Leri, M. Molecular insights into cell toxicity of a novel familial amyloidogenic variant of beta 2-microglobulin.
Li, S. Regulation of the ER stress response by the ion channel activity of the infectious bronchitis Coronavirus envelope protein modulates virion release, apoptosis, viral fitness, and pathogenesis. Ma, J. Jr, Das, S. Nature , 92— Maji, S. Functional amyloids as natural storage of peptide hormones in pituitary secretory granules.
Science , — Morillas, M. Biochemistry 40, — Motta, A. Solution conformation of salmon calcitonin in sodium dodecyl sulfate micelles as determined by two-dimensional NMR and distance geometry calculations. Biochemistry 30, — Nanga, R. Three-dimensional structure and orientation of rat islet amyloid polypeptide protein in a membrane environment by solution NMR spectroscopy.
Natt, N. Nelson, R. Structure of the cross-beta spine of amyloid-like fibrils. Nilsson, L. Alphaantichymotrypsin promotes beta-sheet amyloid plaque deposition in a transgenic mouse model of Alzheimer's disease. Niu, M. RFAmyloid: A web server for predicting amyloid proteins.
Nugent, T. Transmembrane protein topology prediction using support vector machines. Oxana, G. New mechanism of amyloid fibril formation. Current Prot. Pacheco, C. Padmanabhan, J. Alpha1-antichymotrypsin, an inflammatory protein overexpressed in Alzheimer's disease brain, induces tau phosphorylation in neurons. Brain , — Advances in the prediction of protein aggregation propensity. Pasquier, C. Protein Eng. Perissinotto, F. Iron-mediated interaction of alpha synuclein with lipid raft model membranes.
Nanoscale 12, — Press-Sandler, O. Molecular mechanisms of membrane-associated amyloid aggregation: computational perspective and challenges. BBA Biomem. Protter, D. Principles and properties of stress granules. Trends Cell Biol. Pujols, J. Reymer, A. Orientation of aromatic residues in amyloid cores: structural insights into prion fiber diversity. Roy Choudhury, A. Data-driven model for the prediction of protein transmembrane regions.
Amino acid distribution in transmembrane regions: a statistical analysis and comparison with globular proteins. Sawaya, M. Atomic structures of amyloid cross-beta spines reveal varied steric zippers. Sengupta, I. A peptide was predicted as amyloidogenic if it had at least one predicted amyloidogenic region. The presented method named FoldAmyloid allows predicting amyloidogenic regions in a protein or a peptide starting solely from its sequence.
It is based on using expected characteristics—scales: either expected packing density or the probability of formation of hydrogen bonds. The scales themselves were obtained from the statistics of spatial structures of proteins, and then the scales are used for predictions basing on amino acid sequence only.
At first, we obtained these values the expected packing density and probability of formation of hydrogen bonds for each residue in spatial structures of proteins, then calculated average values for each of 20 types of amino acid residues, and then used these average values as the values expected for each residue of a given type in a sequence for which the prediction is made. To analyze the statistics of contacts packing density and hydrogen bonds observed in protein structures, we have constructed Galzitskaya et al.
For each amino acid residue in this database, we have obtained the observed packing density and the observed number of hydrogen bonds.
The observed packing density was calculated Galzitskaya et al. Then, the mean observed packing density for each of the 20 types of amino acid residues was calculated. These mean values are given in our previous work Galzitskaya et al. The mean values were used further for predicting the packing density thus, it is the expected packing density from the protein sequence. Statistics of contacts and hydrogen bonds observed in protein spatial structures.
In b — e , hydrogen bonds are assigned to donor residues gray bars and hydrogen bonds are assigned to acceptor residues black bars.
For backbone-backbone hydrogen bonds, one hydrogen bond per NH-group was considered. For other variants that is, for hydrogen bonds involving side-chains , more than one hydrogen bond for each donor was allowed.
The statistics of hydrogen bonds was analyzed using the same database of three-dimensional protein structures. We searched for four variants of hydrogen bonds separately: backbone-backbone both the donor and the acceptor are in the protein backbone , backbone-side-chain the donor is in the protein backbone while the acceptor is in a side-chain , side-chain-backbone the donor is in a side-chain while the acceptor is in the protein backbone and side-chain-side-chain both the donor and the acceptor are in side-chains.
Then, the probabilities of formation by each type of amino acid residues of hydrogen bonds of a given variant were calculated. Then, this probability was used as a scale for predicting the probability of hydrogen bond formation by each residue in a protein sequence. In Figure 1 b—e , one can see the obtained probabilities of formation of hydrogen bonds of different variants for the 20 types of amino acid residues.
These values will be used further as scales for prediction of amyloidogenic regions. In Table 1 , correlations between some obtained scales are listed. One can see that there is no correlation between the probability of formation of backbone—backbone hydrogen bonds and the probabilities of formation of hydrogen bonds of other types. On the other hand, there is a significant correlation the correlation coefficient is 0.
Correlation coefficients between the scales obtained from the statistics observed in protein structures. The expected values the expected packing density or probability of hydrogen bond formation are used as a scale for constructing a profile the packing density profile or hydrogen bond probability profile, respectively for a protein sequence. First, the corresponding expected value is ascribed to each residue of the sequence it equals to the average value observed for this type of residue in spatial structure ; then, these numbers are averaged inside the window and the average is assigned to the central residue of the window.
On the smoothed profile, we predict a region as amyloidogenic if all its residues lie above the given cutoff in the profile, and the size of the region is greater than or equal to the size of the sliding window used. To evaluate the quality of our predictions as well as to obtain the optimal values for two adjustable parameters the cutoff value and the size of the sliding window for each method based on each scale , we constructed a database of amyloidogenic and non-amyloidogenic peptides by merging the data published by Fernandez-Escamilla et al.
The dataset of Thompson et al. AmylHex is a database of fibril-forming and non-fibril-forming hexapeptides. The dataset of Fernandez-Escamilla et al. The whole merged database consisted of peptides; of these were amyloidogenic while were non-amyloidogenic. We attempted to discriminate between amyloidogenic peptides and non-amyloidogenic ones in this merged database.
A peptide was predicted to be amyloidogenic if at least one amyloidogenic region was predicted in it. The results of the predictions and their comparison with experiment are presented in the form of receiver operator characteristic ROC curves: sensitivity and specificity are plotted for different sliding window sizes different curves at varying cutoff values along the curve.
Figure 2 a represents the results of prediction using the scale of the expected packing density. ROC curves for prediction of amyloidogenic regions in peptides by the FoldAmyloid method with different scales obtained from the statistics of protein structures: a with the scale of expected packing density, b with the scale of probability of formation of backbone-backbone hydrogen bonds donors , c with the scales of probability of formation of backbone-backbone hydrogen bonds acceptors.
For all three scales, the quality of predictions was approximately the same for the sliding window size of three and five amino acid residues. When the size of the sliding window was seven amino acid residues and especially when the size of the sliding window was one amino acid residue that is, without averaging , the quality of predictions was worse see Fig.
Thus, we can use the sliding window size of either three or five residues. Taking into account our previously reported data Galzitskaya et al. Basing on the ROC curves, we selected the cutoff values optimal for amyloidogenic predictions. The optimal vales were the following with the optimal size of the sliding window : 0. Thus, all three scales allow predicting the status of peptides amyloidogenic or non-amyloidogenic with a comparable efficiency. The other three considered scales the probabilities of formation of backbone-side-chain, side-chain-backbone and side-chain-side-chain hydrogen bonds, see Table 1 were unsuccessful in predictions data not shown.
All the obtained ROC-curves lay below the diagonal. That is, non-amyloidogenic peptides in fact have a larger probability to form hydrogen bonds of these types compared to amyloidogenic peptides. Thus, we cannot presently make a large-scale test of the prediction methods for the amyloidogenic fragments of the second type. However, since our method based on the scale of side-chain—side-chain hydrogen bonds finds both amyloidogenic fragments of Sup35 data not shown , we hope that it will work also when more amyloidogenic fragments of the second type are known.
Thus, there are three scales which allow predicting amyloidogenic fragments or rather, the capability of a peptide to be amyloidogenic : the scale of the packing density contact scale , and two scales of the probability to form backbone—backbone hydrogen bonds assigned to donor and to acceptor residues, termed donor and acceptor scales, respectively.
For this purpose, we have normalized each scale to obtain the average over 20 values value equal to zero and standard deviation equal to unity, and then we averaged the normalized scales to obtain a hybrid scale. In Figure 3 , ROC curves for predictions by FoldAmyloid with some hybrid scales are shown the size of the sliding window was five. As shown in the figure, the hybrid scales also give the quality of predictions comparable to that of the two original scales.
ROC curves for prediction of amyloidogenic regions in peptides by the FoldAmyloid method with hybrid scales. The dotted line corresponds to the hybrid scale obtained as a summation of two scales contact and donor scale which were previously normalized the average values were made equal to zero and standard deviations were made equal to unity.
The dashed line corresponds to the hybrid scale obtained by summation of normalized donor scale and acceptor one. The solid line corresponds to the hybrid-scale obtained by summation of all three normalized scales contacts, donors and acceptors. We have compared the quality of predictions by our method with different scales with the predictions of the other methods.
To correctly compare different methods, their performance should be assessed on the same database. Trovato et al. Thus, we predicted amyloidogenic regions separately for this set. The quality of prediction on the same database by our program FoldAmyloid with different scales [the expected packing density scale, the scale of probability of backbone-backbone hydrogen bonds, and the hybrid scale ] as well as by TANGO and PASTA are given in Table 2.
One can see that all analyzed methods are of a comparable efficiency. Performance of different methods of prediction of amyloidogenic regions on a database of peptides.
It should be underlined here the different situation in aggregation process between peptides and globular proteins: in the first case peptide is accessible to the solvent, but in the second case the native state of the globular protein needs to be destabilized in order to the amyloidogenic regions of an amino acid sequence should be exposed to the solvent in order to promote aggregation Tartaglia et al. Therefore, we tried to predict amyloidogenic regions in proteins.
From our previous database of amyloidogenic proteins Galzitskaya et al. In Figure 4 , the results of the predictions are shown for three individual scales which have demonstrated good performance during predictions of amyloidogenic status of peptides.
These are the scale of expected contact density and two scales of probability of backbone—backbone hydrogen bonds: donor and acceptor scale. Gray rectangles indicate positions of experimentally determined amyloidogenic regions in these proteins. The horizontal dashed lines indicate cutoffs.
0コメント