The prediction of main protease SARS-CoV-2 inhibition based on models of enzyme-inhibitor complexes.

A set of linear regression equations predicting the IC50 values for SARS-CoV-2 main protease inhibitors was analyzed. For 180 competitive inhibitors, we have simulated the molecular dynamics of enzyme-inhibitor complexes with known structures or modeled using molecular docking. In the docking procedure, the selection of final poses was restricted by similarity to known structural analogs. The values of the energy contributions obtained by means of calculation of the free energy change of the enzyme-inhibitor complex performed by two variants of the MMPBSA (MMGBSA) method and a number of physicochemical characteristics of the inhibitors were used as independent variables. During the learning process, indicator variables were used for inhibitor subsets obtained from various literature sources to compensate the existing systematic deviations from the target value. A leave one out and leave 20% out cross validation procedures were used to evaluate the prediction quality. For the total logarithmic range width of 3.71, the mean error in predicting the lg(IC50) value was 0.45 log units. The stability of the prediction depending on the variability of the complex in molecular dynamics was investigated.

[The prediction of SARS-CoV-2 main protease inhibition with filtering by position of ligand]. / Predskazanie ingibirovaniia glavnoi proteazy SARS-CoV-2 c uchetom fil'tratsii dannykh o polozhenii ligandov.

The paper analyzes a set of equations that adequately predict the IC50 value for SARS-CoV-2 main protease inhibitors. The training set was obtained using filtering by criteria independent of prediction of target value. It included 76 compounds, and the test set included nine compounds. We used the values of energy contributions obtained in the calculation of the change of the free energy of complex by MMGBSA method and a number of characteristics of the physical and chemical properties of the inhibitors as independent variables. It is sufficient to use only seven independent variables without loss of prediction quality (Q² = 0.79; R²prediction = 0.89). The maximum error in this case does not exceed 0.92 lg(IC50) units with a full range of observed values from 1.26 to 4.95.

[Acetylation, methylation, and ubiquitination of proteins in experimental ischemic stroke in mice: a bioinformatics analysis]. / Atsetilirovanie, metilirovanie i ubikvitinirovanie belkov pri éksperimental'nom ishemicheskom insul'te u myshei: bioinformaticheskii analiz.

The experimental results available in the ProteomeXchange database (accession code PXD016538) (Simats et al. (2020) Molecular and Cellular Proteomics, 19(12), 1921-1936) obtained using a comprehensive multi-omics approach were analyzed in mouse blood to identify potential biomarkers of ischemic stroke. Acetylation, methylation, and ubiquitination were considered as post-translational modifications. The analysis of the significance of changes in the level of protein modification was evaluated for ischemic tissue in comparison with tissue undamaged by stroke and control taken from mice after sham operation. At the level of statistically significant differences according to the Mann-Whitney test (p < 0.05), 2 proteins were found (Q02248 and Q8BL66); for additional 7 proteins, the differences were at the level of a statistical trend (p < 0.1). For 7 of 9 selected proteins there are reports in the literature, for their association with cerebral ischemia.

[The bioinformatic identification of proteins with varying levels of post-translational modifications in experimental ischemic stroke in mice]. / Bioinformaticheskaia identifikatsiia belkov s meniaiushchimsia urovnem posttransliatsionnykh modifikatsii pri éksperimental'nom ishemicheskom insul'te u myshei.

The experimental data obtained by Simats A. et al. (Molecular and Cellular Proteomics, 2020, 19(12), 1921-1936) was analysed using a bioinformatic approach. Original experimental results available in the ProteomeXchange database were obtained using a comprehensive multidomain approach to identify potential blood biomarkers in ischemic stroke in mice. The identification of peptides with post-translational modification (PTM) was performed by us using the raw data (accession code PXD016538). Only phosphorylation and deamination were considered as PTMs. Different combinations of data sets (ischemic tissue with intact tissue, ischemic tissue with control taken from mice after sham surgery, etc.) were compared both in terms of the ratio of abundance for the modified peptide to the unmodified variant and in terms of absolute values of abundance. The most likely change in precisely PTM levels was shown for 27 proteins, which include dynamin, glycogen phosphorylase and 70 kDa heat shock protein.

Prediction of the Inhibition of Influenza Virus Neuraminidase Various Strains by Means of a Generalized Model Constructed Using the Data on the Position of Known Ligands.

Several variants of models for predicting the IC50 values of inhibitors of influenza virus neuraminidase are presented for both individual strains and also for combinations of data for neuraminidases of several strains. They are based on the use of calculated energy contributions to the amount of change in the free energy of enzyme-inhibitor complexes. In contrast to previous works, aimed at the complex modeling, we added a procedure of comparison of the docking variants with one of the neuraminidase inhibitors, for which the structure of the complexes was determined experimentally. Selection of reference molecules for the comparison of structure similarity was made using the Tanimoto metrics and the limit of the RMSD value for a similar part of the structure was no more than 2 Å. Using this limitation and filtering datasets for a particular strain by the Q2 value obtained in the leave-one-out control procedure it was possible to construct equations for predicting the IC50 value with a Q2 value close to the minimum confidence threshold (0.57 in this work). Taking into consideration that in this version of the prediction models, a minimum set of energy contributions is used, which does not employ expensive calculations of entropy contributions, the result obtained supports the correctness of using a generalized model based on the data on the position of known ligands to predict the inhibition of neuraminidase of the influenza virus of various strains.

[Prediction of inhibition of influenza virus neuraminidase of various strains by using a generalized model constructed using the data on the position of known ligands]. / Predskazanie ingibirovaniia neiraminidazy virusa grippa razlichnykh shtammov s ispol'zovaniem obobshchennoi modeli, postroennoi c uchetom dannykh o polozhenii izvestnykh ligandov.

Several variants of models for predicting the IC50 values of inhibitors of influenza virus neuraminidase are presented for both individual strains and for combinations of data for neuraminidases of several strains. They are based on the use of calculated energy contributions to the amount of change in the free energy of enzyme-inhibitor complexes. In contrast to previous works, aimed at the complex modeling, we added a procedure of comparison of the docking variants with one of the neuraminidase inhibitors, for which the structure of the complexes was determined experimentally. The choice of the comparison structure was made according to the similarity of structures evaluated using the Tanimoto metrics and the limit of the RMSD value for a similar part of the structure was no more than 2 Å. Using this limitation and filtering datasets for a particular strain by the Q2 value obtained in the leave-one-out control procedure it is possible to construct equations for predicting the IC50 value with a Q2 value close to the minimum confidence threshold (0.57 in this work). Taking into consideration that in this version of the prediction, a minimum set of energy contributions is used, which does not provide for expensive calculations of entropy contributions, the result obtained supports the correctness of using a generalized model based on the data on the position of known ligands to predict the inhibition of neuraminidase of the influenza virus of various strains.

[The model of prediction of the inhibition of neuraminidases of influenza A and B based on a reduced set of energy terms]. / Model' predskazaniia ingibirovaniia neiraminidaz virusa grippa A i B na baze ogranichennogo nabora énergeticheskikh vkladov.

The overall model for prediction of IC50 values for inhibitors of neuraminidase influenza virus A and B has been created. It combines data about IC50 values of complexes of 40 variants of neuraminidases of influenza A (7 serotypes) and B and three known inhibitors (oseltamivir, zanamivir, peramivir). The model also uses only data of enthalpy contributions to the potential energy of inhibitor/protein and substrate (MUNANA)/protein complexes. The calculation procedures are ported to use software with support of GPU accelerators, that significant decrease the computation time. The corresponding correlation coefficient (R²) for pIC50 prediction was within 0.45-0.58, the SEM values of around 0.7 (the range of used pIC50 data set is from 4.55 to 10.22).

[Creation of a generalized model prediction of inhibition of neuraminidase of influenza virus of various strains]. / Sozdanie obobshchennoi modeli predskazaniia ingibirovaniia neiraminidazy virusa grippa razlichnykh shtammov.

Preliminary results of construction of overall model for prediction of IC50 value of ligands of influenza virus neuraminidase of any strain are presented. We used MM-PBSA (MM-GBSA) energy terms calculated for the complexes obtained after modeling of 30 variants of neuraminidase structures, subsequent docking and simulation of molecular dynamics as independent variables in prediction equations. The structures of known neuraminidase-inhibiting drugs (oseltamivir, zanamivir and peramivir) and a neuraminidase substrate (MUNANA) were used as ligands. The correlation equation based on calculated energetic parameters of inhibitor complexes with neuraminidase did not result in the prediction of IC50 with acceptable parameters (R2£0.3). However, if information about binding energy of the substrate used for neuraminidase assay (and IC50 detection) is included the resulting IC50 prediction equations become significant (R2³0.55). It is concluded that models based on IC50 values as a predictable variable and combining information about binding of different ligands to different variants of the target proteins must take into account the binding properties of the substrate (used for IC50 determination). The predictive power of such models depends critically on the quality of the modeling of the ligand-protein complexes.

[The sequence coverage in different methods of mass spectrometry data analysis obtained on model proteins]. / Stepen' pokrytiia aminokislotnoi posledovatel'nosti pri ispol'zovanii razlichnykh metodov analiza mass-spektrometricheskikh dannykh, poluchennykh na model'nykh belkakh.

The aim of this study was to evaluate sequence coverage of five model proteins (CYB5A, SMAD4, RAB27B, FECH, and CXXC1) by means of shotgun proteomic data analysis employing different methods of data treatment including database-dependent search engines (MASCOT and X!Tandem) and de novo sequencing software ((PEAKS, Novor, and PepNovo+). In order to achieve maximal results, multiprotease hydrolysis including enzymes trypsin, LYS-C, ASPN and GluC was performed in solution and using the FASP method. High resolution mass spectrometry was carried out with a Q EXACTIVE HF hybrid mass spectrometer in the positive ionization mode; parent ions with the highest intensity and a charge range from +2 to +6 were fragmented in the HCD mode. 27 experiments were carried out (hydrolysis with each of 5 enzymes in solution, 4 for the FASP protocol, three technical repeats). Using parameters limiting false identification of peptides, the search engines and de novo sequencing software gave similar results. The degree of sequence coverage was not at least 40%, and in the best cases it reached 80-90%. The use of de novo sequencing software resulted in identification of the Y12H amino acid substitution in one model protein (CYB5A).

[Use of de novo sequencing for proteins identification]. / Primenenie metodov de novo sekvenirovaniia dlia identifikatsii belkov.

Three de novo sequencing programs (Novor, PEAKS and PepNovo+) have been used for identification of 48 individual human proteins constituting the Universal Proteomics Standard Set 2 (UPS2) ("Sigma-Aldrich", USA). Experimental data have been obtained by tandem mass spectrometry. The MS/MS was performed using pure UPS2 and UPS2 mixtures with E. coli extract and human plasma samples. Protein detection was based on identification of at least two peptides of 9 residues in length or one peptide containing at least 13 residues. Using these criteria 13 (Novor), 20 (PEAKS) and 11 (PepNovo+) proteins were detected in pure UPS2 sample. Protein identifications in mixed samples were comparable or worse. Better results (by ~20%) were obtained using prediction included high quality identified fragment (TAG) containing at least 7 residues and unidentified additional masses at N- and C-termini (PepNovo+). The latter approach confidently recognized mass-spectrometric artefacts (and probably PTM). Atypical mass changes missed in UNIMOD DB were found (PepNovo+) to be statistically significant at the C-terminus (+23.02, +26.04 and +27.03). Using peptides containing these modifications and milder detection threshold 41 of 48 UPS2 proteins were identified.

[Correction of the electrophoretic shift in virtual 2D SDS-PAGE electrophoresis]. / Korrektsiia élektroforeticheskogo sdviga v virtual'nom 2D SDS-PAGE élektroforeze.

Virtual electrophoresis in proteomics can be used to search localization of proteins and their proteoforms (especially those existing in low concentrations), to identify proteoforms found in experiments etc. Although the problem of predicting the isoelectric point is well studied, the need of electrophoretic shift correction is usually ignored. Researchers simply use the brutto molecular weight of the protein. In this study four data sets taken from the literature sources and the SWISS-2DPAGE database have been used to build correction equations for prediction of the electrophoretic shift (123, 72, 118 and 470 points, respectively). Two groups of models were built. The first model was based on the amino acid composition of proteins, the second one, on analysis of parameters calculated by amino acid sequences (theoretical molecular weight, hydrophobicity, charge distribution, ability to form helix structures). The coefficient of determination ranged from 0.35 to 0.75 in each single set, but cross-prediction between samples did not gave satisfactory results. At the same time, the direction of correction was predicted correctly in 74% of cases. After combining of the samples and dividing pooled data into 2 representative sets, the coefficient of determination during in the process of learning ranged from 0.44 to 0.51, and R2 of predictions were not less than 0.39. The direction of correction was predicted correctly in 80% of cases. This prediction models have been integrated into the program pIPredict v.2, freely available at http://www.ibmc.msk.ru/LPCIT/pIPredict.

[Prediction of selective inhibition of neuraminidase from various influenza virus strains by potential inhibitors]. / Predskazanie selektivnogo tormozheniia neiraminidazy virusa grippa razlichnykh shtammov potentsial'nymi ingibitorami.

A universal model of inhibition of neuraminidases from various influenza virus strains by a particular has been developed. It is based on known 3D data for neuraminidases from three influenza virus strains (A/Tokyo/3/67, A/tern/Australia/G70C/75, B/Lee/40) and modeling of 3D structure of neuraminidases from other strains (A/PR/8/34 è A/Aichi/2/68). Using docking and molecular dynamics, we have modeled 235 enzyme-ligand complexes for 89 compounds with known IC50 values. Selection of final variants among three results obtained for each enzyme-ligand pair and calculation of independent variables for generation of linear regression equations was performed using MM-PBSA/MM-GBSA. This resulted in the set of equations individual strains and the equations pooling all the data. Thus using this approach it is possible to predict inhibition for neuraminidase from each the considered strains by a particular inhibitor and to predict the range of its action on neuraminidases from various influenza virus strains.

[Cytotoxic activity and molecular modeling of progestins - pregna-D'-pentarans]. / Tsitotoksicheskaia aktivnost' i molekuliarnoe modelirovanie progestinov - pregna-D'-pentaranov.

The cytotoxic activity of synthetic progestins (pregna-D'-pentaranes) II-V full agonists of the progesterone receptor (PR) for PR-positive and PR-negative cells of human breast carcinoma was studied. These compounds were more active in the PR-positive MCF-7 cells than in the PR-negative MDA-MB-453 cells. Cytotoxic effects of tested compounds against normal epithelial MDCK cells were not found. Molecular modeling of studied steroids with PR showed that all progestins with close energy values can bind to the ligand binding domain (LBD) of PR and the magnitude of the energy exceeds the value estimated for the progesterone molecule. Thus, the studied progestins are active against different molecular subtypes of breast cancer and represent a promising class of chemical compounds for oncology.

The mystery of the fourth clone: comparative genomic analysis of four non-typeable Streptococcus pneumoniae strains with different susceptibilities to optochin.

Optochin-resistant pneumococci can be rarely caught in clinical microbiology laboratories because of the routine identification of all such strains as viridans group non-pneumococci. We were lucky to find four non-typeable Streptococcus pneumoniae clones demonstrating the different susceptibilities to optochin: one of them (Spn_13856) was resistant to optochin, while the other three (Spn_1719, Spn_27, and Spn_2298) were susceptible. Whole genome nucleotide sequences of these strains were compared to reveal the differences between the optochin-resistant and optochin-susceptible strains. Two adjacent genes coding maltose O-acetyltransferase and uridine phosphorylase which were presented in the genomes of all optochin-susceptible strains and missed in the optochin-resistant strain were revealed. Non-synonymous substitutions in 14 protein-coding genes were discovered, including the Ala49Ser mutation in the C-subunit of the F0 part of the ATP synthase rotor usually associated with pneumococcal optochin resistance. Modeling of a process of optochin interaction with the F0 part of the ATP synthase rotor indicates that the complex of optochin with "domain C" composed by wild-type C-subunits is more stable than the same complex composed of Ala49Ser mutant C-subunits.

[ProteoСat: a tool for planning of proteomic experiments].

ProteoCat is a computer program has been designed to help researchers in the planning of large-scale proteomic experiments. The central part of this program is the subprogram of hydrolysis simulation that supports 4 proteases (trypsin, lysine C, endoproteinases AspN and GluC). For the peptides obtained after virtual hydrolysis or loaded from data file a number of properties important in mass-spectrometric experiments can be calculated or predicted. The data can be analyzed or filtered to reduce a set of peptides. The program is using new and improved modification of our methods developed to predict pI and probability of peptide detection; pI can also be predicted for a number of popular pKa's scales, proposed by other investigators. The algorithm for prediction of peptide retention time was realized similar to the algorithm used in the program SSRCalc. ProteoCat can estimate the coverage of amino acid sequences of proteins under defined limitation on peptides detection, as well as the possibility of assembly of peptide fragments with user-defined size of "sticky" ends. The program has a graphical user interface, written on JAVA and available at http://www.ibmc.msk.ru/LPCIT/ProteoCat.

[pIPredict: a computer tool for predicting isoelectric points of peptides and proteins].

The data on approximate values of isoelectric point (pI) of peptides obtained during their fractionation by isoelectric focusing can be successfully used for the calculation of the pKa's scale for amino acid residues. This scale can be used for pI prediction. The data of peptide fractionation also provides information about various posttranslational modifications (PTM), so that the prediction of pI may be performed for a wide range of protein forms. In this study, pKa values were calculated using a set of 13448 peptides (including 300 peptides with PTMs significant for pI calculation). The pKa constants were calculated for N-terminal, internal and C-terminal amino acid residues separately. The comparative analysis has shown that our scale increases the accuracy of pI prediction for peptides and proteins and successfully competes with traditional scales and such methods as support vector machines and artificial neural networks. The prediction performed by this scale, can be made in our program pIPredict with GUI written in JAVA as executable jar-archive. The program is freely available for academic users at http://www.ibmc.msk.ru/LPCIT/pIPredict. The software has also the possibility of pI predicting by some other scales; it recognizes some PTM and has the ability to use a custom scale.

[A plain method of prediction of visibility of peptides in mass spectrometry with electrospray ionization].

A new method for screening of essential peptides for protein detection and quantification analysis in the direct positive electrospray mass spectrometry has been proposed. Our method is based on the prediction of the normalized abundance of the mass spectrometric peaks using a linear regression model. This method has the following limitations: (i) selected peptides should be taken so that at pH 2.5 the tested peptides must be presented mainly as the 2+ and 3+ ions; (ii) only peptides having C-terminal lysine or arginine residues are considered. The amino acid composition of the peptide, the peptide concentration, the ratio of the polar surface of peptide to common surface and ratio of the polar volume to common volume are used as independent variables in equation. Several combinations of variables were considered and the best linear regression model had a determination coefficient in leave-one-out validation procedure equal 0.54. This model confidently discriminates peptides with high response ability and peptides with low response ability, and therefore it allows to select only the most promising peptides. This screening method, a plain and fast, can be successfully applied to reduce the list of observed peptides.

[Mode of action prediction of ligands of steroid hormone receptors].

The several predictive models based on two well-known methods PASS and SIMCA were created. These models predict a type of physiological response of steroid compounds binding to nuclear receptors of steroid hormones. We considered 10 variants: the agonists and the antagonists of estrogen, progesterone, androgen, glucocorticoid and mineralocorticoid receptors respectively. Two different sets of descriptors were used during SIMCA (the Dragon descriptors and indices of similarity). The results of discriminant analysis are good enough with average accuracy of 80-85%.

[Molecular docking and 3D-QSAR on 16alpha,17alpha-cycloalkanoprogesterone analogues as progesterone receptor ligands].

A series of 42 steroid ligands was used to predict a binding affinity to progesterone receptor. The molecules were the derivatives of 16alpha,17alpha-cycloalkanoprogesterones. Different methods of prediction were used and analyzed such as CoMFA and artificial neural networks. The best result (Q2 = 0.91) was obtained for a combination of molecular docking, molecular dynamics simulation and artificial neural networks. A predictive power of the model was validated by a group of 8 pentarans synthesized separately and tested in vitro (R2test = 0.77). This model can be used to determine the affinity level of the ligand to progesterone receptor and accurate ranking of binding compounds.

[Molecular modeling of acetylcholinesterase interaction with irreversible and reversible organophosphorous inhibitors].

Three-dimensional Quantitative Structure-Activity Relationship models were designed for irreversible and reversible acetylcholinesterase inhibitors by molecular modeling methods. In case of irreversible inhibitors CoMFA (the comparative analysis of molecular fields) or CoMSIA (the comparative analysis of indexes of molecular similarity) descriptors together with HYBOT 3D fields provide more statistically valid 3D-QSAR models. This indicates importance of donor-acceptor interactions for irreversible acetylcholinesterase inhibition. In case of reversible organophosphorous inhibitors good quality model for structure-activity relationships was developed using CoMFA fields. The obtained models have good predictive power and can be used for estimation of new organophosphorous compounds inhibitor activity that in turn correlates with toxicity of these compounds.