Comprehensive classification and diversity assessment of atomic contacts in protein-small ligand interactions

E2EATP: Fast and High-Accuracy Protein-ATP Binding Residue Prediction via Protein Language Model Embedding

Identifying the ATP-binding sites of proteins is fundamentally important to uncover the mechanisms of protein functions and explore drug discovery. Many computational methods are proposed to predict ATP-binding sites. However, due to the limitation of the quality of feature representation, the prediction performance still has a big room for improvement. In this study, we propose an end-to-end deep learning model, E2EATP, to dig out more discriminative information from a protein sequence for improving the ATP-binding site prediction performance. Concretely, we employ a pretrained deep learning-based protein language model (ESM2) to automatically extract high-latent discriminative representations of protein sequences relevant for protein functions. Based on ESM2, we design a residual convolutional neural network to train a protein-ATP binding site prediction model. Furthermore, a weighted focal loss function is used to reduce the negative impact of imbalanced data on the model training stage. Experimental results on the two independent testing data sets demonstrate that E2EATP could achieve higher Matthew's correlation coefficient and AUC values than most existing state-of-the-art prediction methods. The speed (about 0.05 s per protein) of E2EATP is much faster than the other existing prediction methods. Detailed data analyses show that the major advantage of E2EATP lies at the utilization of the pretrained protein language model that extracts more discriminative information from the protein sequence only. The standalone package of E2EATP is freely available for academic at https://github.com/jun-csbio/e2eatp/.

Inverse Mixed-Solvent Molecular Dynamics for Visualization of the Residue Interaction Profile of Molecular Probes

To ensure efficiency in discovery and development, the application of computational technology is essential. Although virtual screening techniques are widely applied in the early stages of drug discovery research, the computational methods used in lead optimization to improve activity and reduce the toxicity of compounds are still evolving. In this study, we propose a method to construct the residue interaction profile of the chemical structure used in the lead optimization by performing “inverse” mixed-solvent molecular dynamics (MSMD) simulation. Contrary to constructing a protein-based, atom interaction profile, we constructed a probe-based, protein residue interaction profile using MSMD trajectories. It provides us the profile of the preferred protein environments of probes without co-crystallized structures. We assessed the method using three probes: benzamidine, catechol, and benzene. As a result, the residue interaction profile of each probe obtained by MSMD was a reasonable physicochemical description of the general non-covalent interaction. Moreover, comparison with the X-ray structure containing each probe as a ligand shows that the map of the interaction profile matches the arrangement of amino acid residues in the X-ray structure.

Crystallography-based exploration of non-covalent interactions for the design and synthesis of coumarin for stronger protein binding

Protein molecules are a good target for the inhibition or promotion of biological processes. Different methods like QSAR and molecular docking have been developed to accurately design small binder molecules for target proteins. An alternative model has been developed wherein a statistical method is used to find the propensity of different non-covalent interactions between small molecules and amino acid residues of the protein. The results give hints as to the choice of substituents required at the SM to strongly bind to a protein. In this case, 75 different types of proteins bound with coumarin derivatives have been investigated and the non-covalent interactions observed between the basic coumarin moiety and amino acids have been analyzed. Density functional theory (DFT) calculations were used to identify the electronic features of coumarin to understand the feasibility of the observed non-covalent interactions and to find appropriate groups that can modulate these interactions. The binding affinity towards a protein (β-lactoglobulin (BLG)) and the stability of the protein complex have been investigated through docking and molecular dynamics of 100 ns, respectively. The modeled compounds were synthesized and investigated with regards to their interactions with the model carrier protein. The thermodynamics of the interactions were also investigated and the binding is governed by the Le Chatelier principle.

A crystallography-based investigation of weak interactions for drug design against COVID-19

Interactions between proteins and small molecules play important roles in the inhibition of protein function. However, a lack of proper knowledge about non-covalent interactions can act as a barrier towards gaining a complete understanding of the factors that control these associations. To find effective molecules for COVID-19 inhibition, we have quantitatively investigated 143 X-ray crystal structures of the SARS-CoV-2 M^pro protein of coronavirus with covalently or non-covalently bound small molecules (SMs). Our present study is able to explain ordinary and perceptive aspects relating to protein inhibition. The active site of the protein consists of 21 amino acid residues, but only nine are actively involved in the ligand binding process. The H41, M49, and C145 residues have highest priority with respect to interactions with small molecules through hydrogen bond, CH-π, and van der Waals interactions. At the active site, this ranking of amino acids is clear, based on different spatial orientations of ligands, and consistent with the electronic properties. SMs with aromatic moieties that bind to the active site of the protein play a distinct role in the determination of the following order of interaction frequency with the amino acids: CH-π > H-bonding > polar interactions. This present study revealed that the G143 and C145 residues play crucial roles in the recognition of the carbonyl functionality of SMs through hydrogen bonding. With this knowledge in mind, an effective inhibitor small-molecule for SARS-CoV-2 M^pro was designed: docking studies showed that the designed molecule has strong binding affinity towards the protein. The non-covalent interactions in the protein-ligand complex are in good agreement with the results obtained from X-ray crystallography. Moreover, the present study focused on weak forces and their influence on protein inhibition, henceforth shedding much light on the essential requirements for moieties that should be present in a good inhibitor and their orientations at the ligand binding site.

Sequence-based prediction of physicochemical interactions at protein functional sites using a function-and-interaction-annotated domain profile database

Background

Identifying protein functional sites (PFSs) and, particularly, the physicochemical interactions at these sites is critical to understanding protein functions and the biochemical reactions involved. Several knowledge-based methods have been developed for the prediction of PFSs; however, accurate methods for predicting the physicochemical interactions associated with PFSs are still lacking.

Results

In this paper, we present a sequence-based method for the prediction of physicochemical interactions at PFSs. The method is based on a functional site and physicochemical interaction-annotated domain profile database, called fiDPD, which was built using protein domains found in the Protein Data Bank. This method was applied to 13 target proteins from the very recent Critical Assessment of Structure Prediction (CASP10/11), and our calculations gave a Matthews correlation coefficient (MCC) value of 0.66 for PFS prediction and an 80% recall in the prediction of the associated physicochemical interactions.

Conclusions

Our results show that, in addition to the PFSs, the physical interactions at these sites are also conserved in the evolution of proteins. This work provides a valuable sequence-based tool for rational drug design and side-effect assessment. The method is freely available and can be accessed at http://202.119.249.49.

ATPbind: Accurate Protein-ATP Binding Site Prediction by Combining Sequence-Profiling and Structure-Based Comparisons

Protein-ATP interactions are ubiquitous in a wide variety of biological processes. Correctly locating ATP binding sites from protein information is an important but challenging task for protein function annotation and drug discovery. However, there is no method that can optimally identify ATP binding sites for different proteins. In this study, we report a new composite predictor, ATPbind, for ATP binding sites by integrating the outputs of two template-based predictors (i.e., S-SITE and TM-SITE) and three discriminative sequence-driven features of proteins: position specific scoring matrix, predicted secondary structure, and predicted solvent accessibility. In ATPbind, we assembled multiple support vector machines (SVMs) based on a random undersampling technique to cope with the serious imbalance phenomenon between the numbers of ATP binding sites and of non-ATP binding sites. We also constructed a new gold-standard benchmark data set consisting of 429 ATP binding proteins from the PDB database to evaluate and compare the proposed ATPbind with other existing predictors. Starting from a query sequence and predicted I-TASSER models, ATPbind can achieve an average accuracy of 72%, covering 62% of all ATP binding sites while achieving a Matthews correlation coefficient value that is significantly higher than that of other state-of-the-art predictors.

Enhance the performance of current scoring functions with the aid of 3D protein-ligand interaction fingerprints

BACKGROUND

In structure-based drug design, binding affinity prediction remains as a challenging goal for current scoring functions. Development of target-biased scoring functions provides a new possibility for tackling this problem, but this approach is also associated with certain technical difficulties. We previously reported the Knowledge-Guided Scoring (KGS) method as an alternative approach (BMC Bioinformatics, 2010, 11, 193-208). The key idea is to compute the binding affinity of a given protein-ligand complex based on the known binding data of an appropriate reference complex, so the error in binding affinity prediction can be reduced effectively.

RESULTS

In this study, we have developed an upgraded version, i.e. KGS2, by employing 3D protein-ligand interaction fingerprints in reference selection. KGS2 was evaluated in combination with four scoring functions (X-Score, ChemPLP, ASP, and GoldScore) on five drug targets (HIV-1 protease, carbonic anhydrase 2, beta-secretase 1, beta-trypsin, and checkpoint kinase 1). In the in situ scoring test, considerable improvements were observed in most cases after application of KGS2. Besides, the performance of KGS2 was always better than KGS in all cases. In the more challenging molecular docking test, application of KGS2 also led to improved structure-activity relationship in some cases.

CONCLUSIONS

KGS2 can be applied as a convenient "add-on" to current scoring functions without the need to re-engineer them, and its application is not limited to certain target proteins as customized scoring functions. As an interpolation method, its accuracy in principle can be improved further with the increasing knowledge of protein-ligand complex structures and binding affinity data. We expect that KGS2 will become a practical tool for enhancing the performance of current scoring functions in binding affinity prediction. The KGS2 software is available upon contacting the authors.

Landscape of protein-small ligand binding modes

Elucidating the mechanisms of specific small-molecule (ligand) recognition by proteins is a long-standing conundrum. While the structures of these molecules, proteins and ligands, have been extensively studied, protein-ligand interactions, or binding modes, have not been comprehensively analyzed. Although methods for assessing similarities of binding site structures have been extensively developed, the methods for the computational treatment of binding modes have not been well established. Here, we developed a computational method for encoding the information about binding modes as graphs, and assessing their similarities. An all-against-all comparison of 20,040 protein-ligand complexes provided the landscape of the protein-ligand binding modes and its relationships with protein- and chemical spaces. While similar proteins in the same SCOP Family tend to bind relatively similar ligands with similar binding modes, the correlation between ligand and binding similarities was not very high (R(2)  = 0.443). We found many pairs with novel relationships, in which two evolutionally distant proteins recognize dissimilar ligands by similar binding modes (757,474 pairs out of 200,790,780 pairs were categorized into this relationship, in our dataset). In addition, there were an abundance of pairs of homologous proteins binding to similar ligands with different binding modes (68,217 pairs). Our results showed that many interesting relationships between protein-ligand complexes are still hidden in the structure database, and our new method for assessing binding mode similarities is effective to find them.

A novel approach of dynamic cross correlation analysis on molecular dynamics simulations and its application to Ets1 dimer-DNA complex

The dynamic cross correlation (DCC) analysis is a popular method for analyzing the trajectories of molecular dynamics (MD) simulations. However, it is difficult to detect correlative motions that appear transiently in only a part of the trajectory, such as atomic contacts between the side-chains of amino acids, which may rapidly flip. In order to capture these multi-modal behaviors of atoms, which often play essential roles, particularly at the interfaces of macromolecules, we have developed the "multi-modal DCC (mDCC)" analysis. The mDCC is an extension of the DCC and it takes advantage of a Bayesian-based pattern recognition technique. We performed MD simulations for molecular systems modeled from the (Ets1)2-DNA complex and analyzed their results with the mDCC method. Ets1 is an essential transcription factor for a variety of physiological processes, such as immunity and cancer development. Although many structural and biochemical studies have so far been performed, its DNA binding properties are still not well characterized. In particular, it is not straightforward to understand the molecular mechanisms how the cooperative binding of two Ets1 molecules facilitates their recognition of Stromelysin-1 gene regulatory elements. A correlation network was constructed among the essential atomic contacts, and the two major pathways by which the two Ets1 molecules communicate were identified. One is a pathway via direct protein-protein interactions and the other is that via the bound DNA intervening two recognition helices. These two pathways intersected at the particular cytosine bases (C110/C11), interacting with the H1, H2, and H3 helices. Furthermore, the mDCC analysis showed that both pathways included the transient interactions at their intermolecular interfaces of Tyr396-C11 and Ala327-Asn380 in multi-modal motions of the amino acid side chains and the nucleotide backbone. Thus, the current mDCC approach is a powerful tool to reveal these complicated behaviors and scrutinize intermolecular communications in a molecular system.

C-H…O hydrogen bonds in FK506-binding protein-ligand interactions

Hydrogen bonds are important interaction forces observed in protein structures. They can be classified as stronger or weaker depending on their energy, thereby reflecting on the type of donor. The contribution of weak hydrogen bonds is deemed as an important factor toward structure stability along with the stronger bonds. One such bond, the C-H…O type hydrogen bond, is shown to make a contribution in maintaining three dimensional structures of proteins. Apart from their presence within protein structures, the role of these bonds in protein-ligand interactions is also noteworthy. In this study, we present a statistical analysis on the presence of C-H…O hydrogen bonds observed between FKBPs and their cognate ligands. The FK506-binding proteins (FKBPs) carry peptidyl cis-trans isomerase activity apart from the immunosuppressive property by binding to the immunosuppressive drugs FK506 or rapamycin. Because the active site of FKBPs is lined up by many hydrophobic residues, we speculated that the prevalence of C-H…O hydrogen bonds will be considerable. In a total of 25 structures analyzed, a higher frequency of C-H…O hydrogen bonds is observed in comparison with the stronger hydrogen bonds. These C-H…O hydrogen bonds are dominated by a highly conserved donor, the C(α/β) of Val55 and an acceptor, the backbone oxygen of Glu54. Both these residues are positioned in the β4-α1 loop, whereas the other residues Tyr26, Phe36 and Phe99 with higher frequencies are lined up at the opposite face of the active site. These preferences could be implicated in FKBP pharmacophore models toward enhancing the ligand affinity. This study could be a prelude to studying other proteins with hydrophobic pockets to gain better insights into ligand recognition.

Polypharmacology rescored: protein-ligand interaction profiles for remote binding site similarity assessment

Detection of remote binding site similarity in proteins plays an important role for drug repositioning and off-target effect prediction. Various non-covalent interactions such as hydrogen bonds and van-der-Waals forces drive ligands' molecular recognition by binding sites in proteins. The increasing amount of available structures of protein-small molecule complexes enabled the development of comparative approaches. Several methods have been developed to characterize and compare protein-ligand interaction patterns. Usually implemented as fingerprints, these are mainly used for post processing docking scores and (off-)target prediction. In the latter application, interaction profiles detect similarities in the bound interactions of different ligands and thus identify essential interactions between a protein and its small molecule ligands. Interaction pattern similarity correlates with binding site similarity and is thus contributing to a higher precision in binding site similarity assessment of proteins with distinct global structure. This renders it valuable for existing drug repositioning approaches in structural bioinformatics. Current methods to characterize and compare structure-based interaction patterns - both for protein-small-molecule and protein-protein interactions - as well as their potential in target prediction will be reviewed in this article. The question of how the set of interaction types, flexibility or water-mediated interactions, influence the comparison of interaction patterns will be discussed. Due to the wealth of protein-ligand structures available today, predicted targets can be ranked by comparing their ligand interaction pattern to patterns of the known target. Such knowledge-based methods offer high precision in comparison to methods comparing whole binding sites based on shape and amino acid physicochemical similarity.

GIANT: pattern analysis of molecular interactions in 3D structures of protein-small ligand complexes

Background

Interpretation of binding modes of protein–small ligand complexes from 3D structure data is essential for understanding selective ligand recognition by proteins. It is often performed by visual inspection and sometimes largely depends on a priori knowledge about typical interactions such as hydrogen bonds and π-π stacking. Because it can introduce some biases due to scientists’ subjective perspectives, more objective viewpoints considering a wide range of interactions are required.

Description

In this paper, we present a web server for analyzing protein–small ligand interactions on the basis of patterns of atomic contacts, or “interaction patterns” obtained from the statistical analyses of 3D structures of protein–ligand complexes in our previous study. This server can guide visual inspection by providing information about interaction patterns for each atomic contact in 3D structures. Users can visually investigate what atomic contacts in user-specified 3D structures of protein–small ligand complexes are statistically overrepresented. This server consists of two main components: “Complex Analyzer”, and “Pattern Viewer”. The former provides a 3D structure viewer with annotations of interacting amino acid residues, ligand atoms, and interacting pairs of these. In the annotations of interacting pairs, assignment to an interaction pattern of each contact and statistical preferences of the patterns are presented. The “Pattern Viewer” provides details of each interaction pattern. Users can see visual representations of probability density functions of interactions, and a list of protein–ligand complexes showing similar interactions.

Conclusions

Users can interactively analyze protein–small ligand binding modes with statistically determined interaction patterns rather than relying on a priori knowledge of the users, by using our new web server named GIANT that is freely available at http://giant.hgc.jp/.