AK-Score: Accurate Protein-Ligand Binding Affinity Prediction Using an Ensemble of 3D-Convolutional Neural Networks

Accurate prediction of the binding affinity of a protein-ligand complex is essential for efficient and successful rational drug design. Therefore, many binding affinity prediction methods have been developed. In recent years, since deep learning technology has become powerful, it is also implemented to predict affinity. In this work, a new neural network model that predicts the binding affinity of a protein-ligand complex structure is developed. Our model predicts the binding affinity of a complex using the ensemble of multiple independently trained networks that consist of multiple channels of 3-D convolutional neural network layers. Our model was trained using the 3772 protein-ligand complexes from the refined set of the PDBbind-2016 database and tested using the core set of 285 complexes. The benchmark results show that the Pearson correlation coefficient between the predicted binding affinities by our model and the experimental data is 0.827, which is higher than the state-of-the-art binding affinity prediction scoring functions. Additionally, our method ranks the relative binding affinities of possible multiple binders of a protein quite accurately, comparable to the other scoring functions. Last, we measured which structural information is critical for predicting binding affinity and found that the complementarity between the protein and ligand is most important.     

Valproic Acid as a Potential Inhibitor of Plasmodium falciparum Histone Deacetylase 1 (PfHDAC1): An in Silico Approach

A new Plasmodium falciparum histone deacetylase1 (PfHDAC1) homology model was built based on the highest sequence identity available template human histone deacetylase 2 structure. The generated model was carefully evaluated for stereochemical accuracy, folding correctness and overall structure quality. All evaluations were acceptable and consistent. Docking a group of hydroxamic acid histone deacetylase inhibitors and valproic acid has shown binding poses that agree well with inhibitor-bound histone deacetylase-solved structural interactions. Docking affinity dG scores were in agreement with available experimental binding affinities. Further, enzyme-ligand complex stability and reliability were investigated by running 5-nanosecond molecular dynamics simulations. Thorough analysis of the simulation trajectories has shown that enzyme-ligand complexes were stable during the simulation period. Interestingly, the calculated theoretical binding energies of the docked hydroxamic acid inhibitors have shown that the model can discriminate between strong and weaker inhibitors and agrees well with the experimental affinities reported in the literature. The model and the docking methodology can be used in screening virtual libraries for PfHDAC1 inhibitors, since the docking scores have ranked ligands in accordance with experimental binding affinities. Valproic acid calculated theoretical binding energy suggests that it may inhibit PfHDAC1.     

Towards Automated Binding Affinity Prediction Using an Iterative Linear Interaction Energy Approach

Binding affinity prediction of potential drugs to target and off-target proteins is an essential asset in drug development. These predictions require the calculation of binding free energies. In such calculations, it is a major challenge to properly account for both the dynamic nature of the protein and the possible variety of ligand-binding orientations, while keeping computational costs tractable. Recently, an iterative Linear Interaction Energy (LIE) approach was introduced, in which results from multiple simulations of a protein-ligand complex are combined into a single binding free energy using a Boltzmann weighting-based scheme. This method was shown to reach experimental accuracy for flexible proteins while retaining the computational efficiency of the general LIE approach. Here, we show that the iterative LIE approach can be used to predict binding affinities in an automated way. A workflow was designed using preselected protein conformations, automated ligand docking and clustering, and a (semi-)automated molecular dynamics simulation setup. We show that using this workflow, binding affinities of aryloxypropanolamines to the malleable Cytochrome P450 2D6 enzyme can be predicted without a priori knowledge of dominant protein-ligand conformations. In addition, we provide an outlook for an approach to assess the quality of the LIE predictions, based on simulation outcomes only.     

Artificial Intelligence in Aptamer–Target Binding Prediction

Aptamers are short single-stranded DNA, RNA, or synthetic Xeno nucleic acids (XNA) molecules that can interact with corresponding targets with high affinity. Owing to their unique features, including low cost of production, easy chemical modification, high thermal stability, reproducibility, as well as low levels of immunogenicity and toxicity, aptamers can be used as an alternative to antibodies in diagnostics and therapeutics. Systematic evolution of ligands by exponential enrichment (SELEX), an experimental approach for aptamer screening, allows the selection and identification of in vitro aptamers with high affinity and specificity. However, the SELEX process is time consuming and characterization of the representative aptamer candidates from SELEX is rather laborious. Artificial intelligence (AI) could help to rapidly identify the potential aptamer candidates from a vast number of sequences. This review discusses the advancements of AI pipelines/methods, including structure-based and machine/deep learning-based methods, for predicting the binding ability of aptamers to targets. Structure-based methods are the most used in computer-aided drug design. For this part, we review the secondary and tertiary structure prediction methods for aptamers, molecular docking, as well as molecular dynamic simulation methods for aptamer–target binding. We also performed analysis to compare the accuracy of different secondary and tertiary structure prediction methods for aptamers. On the other hand, advanced machine-/deep-learning models have witnessed successes in predicting the binding abilities between targets and ligands in drug discovery and thus potentially offer a robust and accurate approach to predict the binding between aptamers and targets. The research utilizing machine-/deep-learning techniques for prediction of aptamer–target binding is limited currently. Therefore, perspectives for models, algorithms, and implementation strategies of machine/deep learning-based methods are discussed. This review could facilitate the development and application of high-throughput and less laborious in silico methods in aptamer selection and characterization.     

From knowledge-based potentials to combinatorial lead design in silico

Computational methods are becoming increasingly used in the drug discovery process. In this Account, we review a novel computational method for lead discovery. This method, called CombiSMoG for "combinatorial small molecule growth", is based on two components: a fast and accurate knowledge-based scoring function used to predict binding affinities of protein-ligand complexes, and a Monte Carlo combinatorial growth algorithm that generates large numbers of low-free-energy ligands in the binding site of a protein. We illustrate the advantages of the method by describing its application in the design of picomolar inhibitors for human carbonic anhydrase.     

Incorporating intermolecular distance into protein-protein docking

In this work, intermolecular distance was integrated into the docking of protein-protein complexes. To develop an efficient docking procedure, 22 enzyme-inhibitor targets and 15 antibody-antigen targets were taken from a benchmark set. A three-step approach was adopted, which included global sampling by FTDOCK, filtering by intermolecular distance and ranking by a composite scoring function. For the enzyme-inhibitor targets, the composite scoring function consists of geometry and energy terms. In the set composed of the approximately 100 highest ranked candidates for each target, correct complexes were identified for all of the 22 enzyme-inhibitor targets. This docking strategy also succeeded on the four test targets, of which three are CAPRI targets with the same receptor but different binding modes. Interestingly, all three binding modes were correctly predicted. For the antibody-antigen targets, CDR and physical energy were also used in the filtering process and informatics terms were added to the scoring function. The composite score had successful prediction for 13 of the 15 antibody-antigen targets.     

DockingApp RF: A State-of-the-Art Novel Scoring Function for Molecular Docking in a User-Friendly Interface to AutoDock Vina

Motivation: Bringing a new drug to the market is expensive and time-consuming. To cut the costs and time, computer-aided drug design (CADD) approaches have been increasingly included in the drug discovery pipeline. However, despite traditional docking tools show a good conformational space sampling ability, they are still unable to produce accurate binding affinity predictions. This work presents a novel scoring function for molecular docking seamlessly integrated into DockingApp, a user-friendly graphical interface for AutoDock Vina. The proposed function is based on a random forest model and a selection of specific features to overcome the existing limits of Vina’s original scoring mechanism. A novel version of DockingApp, named DockingApp RF, has been developed to host the proposed scoring function and to automatize the rescoring procedure of the output of AutoDock Vina, even to nonexpert users. Results: By coupling intermolecular interaction, solvent accessible surface area features and Vina’s energy terms, DockingApp RF’s new scoring function is able to improve the binding affinity prediction of AutoDock Vina. Furthermore, comparison tests carried out on the CASF-2013 and CASF-2016 datasets demonstrate that DockingApp RF’s performance is comparable to other state-of-the-art machine-learning- and deep-learning-based scoring functions. The new scoring function thus represents a significant advancement in terms of the reliability and effectiveness of docking compared to AutoDock Vina’s scoring function. At the same time, the characteristics that made DockingApp appealing to a wide range of users are retained in this new version and have been complemented with additional features.     

Trastuzumab-Peptide Interactions: Mechanism and Application in Structure-Based Ligand Design

Understanding of protein-ligand interactions and its influences on protein stability is necessary in the research on all biological processes and correlative applications, for instance, the appropriate affinity ligand design for the purification of bio-drugs. In this study, computational methods were applied to identify binding site interaction details between trastuzumab and its natural receptor. Trastuzumab is an approved antibody used in the treatment of human breast cancer for patients whose tumors overexpress the HER2 (human epidermal growth factor receptor 2) protein. However, rational design of affinity ligands to keep the stability of protein during the binding process is still a challenge. Herein, molecular simulations and quantum mechanics were used on protein-ligand interaction analysis and protein ligand design. We analyzed the structure of the HER2-trastuzumab complex by molecular dynamics (MD) simulations. The interaction energies of the mutated peptides indicate that trastuzumab binds to ligand through electrostatic and hydrophobic interactions. Quantitative investigation of interactions shows that electrostatic interactions play the most important role in the binding of the peptide ligand. Prime/MM-GBSA calculations were carried out to predict the binding affinity of the designed peptide ligands. A high binding affinity and specificity peptide ligand is designed rationally with equivalent interaction energy to the wild-type octadecapeptide. The results offer new insights into affinity ligand design.     

Limitations and Extensions of the Lock-and-Key Principle: Differences between Gas State,Solution and Solid State Structures

The lock-and-key concept is discussed with respect to necessary extensions. Formation of supramolecular complexes depends not only, and often not even primarily on an optimal geometric fit between host and guest. Induced fit and allosteric interactions have long been known as important modifications. Different binding mechanisms, the medium used and pH effects can exert a major influence on the affinity. Stereoelectronic effects due to lone pair orientation can lead to variation of binding constants by orders of magnitude. Hydrophobic interactions due to high-energy water inside cavities modify the mechanical lock-and-key picture. That optimal affinities are observed if the cavity is only partially filled by the ligand can be in conflict with the lock-and-key principle. In crystals other forces than those between host and guest often dominate, leading to differences between solid state and solution structures. This is exemplified in particular with calixarene complexes, which by X-ray analysis more often than other hosts show guest molecules outside their cavity. In view of this the particular problems with the identification of weak interactions in crystals is discussed.     

Advances and challenges in protein-ligand docking

Molecular docking is a widely-used computational tool for the study of molecular recognition, which aims to predict the binding mode and binding affinity of a complex formed by two or more constituent molecules with known structures. An important type of molecular docking is protein-ligand docking because of its therapeutic applications in modern structure-based drug design. Here, we review the recent advances of protein flexibility, ligand sampling, and scoring functions-the three important aspects in protein-ligand docking. Challenges and possible future directions are discussed in the Conclusion.     

A Fragment-Based Scoring Function to Re-rank ATP Docking Results

ATP is involved in numerous biochemical reactions in living cells interacting with different proteins. Molecular docking simulations provide considerable insight into the problem of molecular recognition of this substrate. To improve the selection of correct ATP poses among those generated by docking algorithms we propose a post-docking reranking criterion. The method is based on detailed analysis of the intermolecular interactions in 50 high-resolution 3D-structures of ATP-protein complexes. A distinctive new feature of the proposed method is that the ligand molecule is divided into fragments that differ in their physical properties. The placement of each of them into the binding site is judged separately by different criteria, thus avoiding undesirable averaging of the scoring function terms by highlighting those relevant for particular fragments. The scoring performance of the new criteria was tested with the docking solutions for ATPprotein complexes and a significant improvement in the selection of correct docking poses was observed, as compared to the standard scoring function.     

Virtual Screening of Specific Insulin-Like Growth Factor 1 Receptor (IGF1R) Inhibitors from the National Cancer Institute (NCI) Molecular Database

Insulin-like growth factor 1 receptor (IGF1R) is an attractive drug target for cancer therapy and research on IGF1R inhibitors has had success in clinical trials. A particular challenge in the development of specific IGF1R inhibitors is interference from insulin receptor (IR), which has a nearly identical sequence. A few potent inhibitors that are selective for IGF1R have been discovered experimentally with the aid of computational methods. However, studies on the rapid identification of IGF1R-selective inhibitors using virtual screening and confidence-level inspections of ligands that show different interactions with IGF1R and IR in docking analysis are rare. In this study, we established virtual screening and binding-mode prediction workflows based on benchmark results of IGF1R and several kinase receptors with IGF1R-like structures. We used comprehensive analysis of the known complexes of IGF1R and IR with their binding ligands to screen specific IGF1R inhibitors. Using these workflows, 17 of 139,735 compounds in the NCI (National Cancer Institute) database were identified as potential specific inhibitors of IGF1R. Calculations of the potential of mean force (PMF) with GROMACS were further conducted for three of the identified compounds to assess their binding affinity differences towards IGF1R and IR.     

NMR‐Guided Molecular Docking of a Protein–Peptide Complex Based on Ant Colony Optimization

Standard docking approaches used for the prediction of protein–ligand complexes in the drug development process have problems identifying the correct binding mode of large flexible ligands. Herein we show how additional experimental data from NMR experiments can be used to predict the binding mode of a mucin 1 (MUC‐1) pentapeptide recognized by the breast‐cancer‐selective monoclonal antibody SM3. Distance constraints derived from trNOE and saturation transfer difference NMR experiments are combined with the docking approach PLANTS. The resulting complex structures show excellent agreement with the NMR data and with a published X‐ray crystal structure. The method was then further tested on two complexes in order to demonstrate its more general applicability: T‐antigen disaccharide bound to Maclura pomifera agglutinin, and the inhibitor SBi279 bound to S100B protein. Our new approach has the advantages of being fully automatic, rapid, and unbiased; moreover, it is based on relatively easily obtainable experimental data and can greatly increase the reliability of the generated structures.     

An experimental and molecular-modeling study of the binding of linked sulfated tetracyclitols to FGF-1 and FGF-2

The experimental binding affinities of a series of linked sulfated tetracyclitols [Cyc2N-R-NCyc2, where Cyc = C6H6(OSO3Na)3 and R = (CH2)n (n = 2-10), p-xylyl or (C2H4)2-Ncyc] for the fibroblast growth factors FGF-1 and FGF-2 have been measured by using a surface plasmon resonance assay. The KD values range from 7.0 nM to 1.1 microM for the alkyl-linked ligands. The binding affinity is independent of the flexibility of the linker, as replacement of the alkyl linker with a rigid p-xylyl group did not affect the KD. Calculations suggest that binding modes for the p-xylyl-linked ligand are similar to those calculated for the flexible alkyl-linked tetracyclitols. The possible formation of cross-linked FGF:cyclitol complexes was examined by determining KD values at increasing protein concentrations. No changes in KD were observed; this suggesting that only 1:1 complexes are formed under these assay conditions. Monte Carlo multiple-minima calculations of low-energy conformers of the FGF-bound ligands showed that all of the sulfated tetracyclitol ligands can bind effectively in the heparan sulfate-binding sites of FGF-1 and FGF-2. Binding affinities of these complexes were estimated by the Linear Interaction Energy (LIE) method to within a root-mean-square deviation of 1 kcal mol(-1) of the observed values. The effect of incorporating cations to balance the overall charge of the complexes during the LIE calculations was also explored.     

Methods and Applications of In Silico Aptamer Design and Modeling

Aptamers are nucleic acid analogues of antibodies with high affinity to different targets, such as cells, viruses, proteins, inorganic materials, and coenzymes. Empirical approaches allow the design of in vitro aptamers that bind particularly to a target molecule with high affinity and selectivity. Theoretical methods allow significant expansion of the possibilities of aptamer design. In this study, we review theoretical and joint theoretical-experimental studies dedicated to aptamer design and modeling. We consider aptamers with different targets, such as proteins, antibiotics, organophosphates, nucleobases, amino acids, and drugs. During nucleic acid modeling and in silico design, a full set of in silico methods can be applied, such as docking, molecular dynamics (MD), and statistical analysis. The typical modeling workflow starts with structure prediction. Then, docking of target and aptamer is performed. Next, MD simulations are performed, which allows for an evaluation of the stability of aptamer/ligand complexes and determination of the binding energies with higher accuracy. Then, aptamer/ligand interactions are analyzed, and mutations of studied aptamers made. Subsequently, the whole procedure of molecular modeling can be reiterated. Thus, the interactions between aptamers and their ligands are complex and difficult to understand using only experimental approaches. Docking and MD are irreplaceable when aptamers are studied in silico.     

Pharmacological Profile and Molecular Modeling of Cyclic Opioid Analogues Incorporating Various Phenylalanine Derivatives

Peptide-based agonists of the μ opioid receptor (μOR) are promising therapeutic candidates for pain relief with reduced side effects compared to morphine. A deep understanding of μOR–ligand interactions is necessary for future design of peptide-based opioid analgesics. To explore the requirements of the μOR binding pocket, eight new analogues of our cyclic peptide Tyr-c[d -Lys−Phe−Phe−Asp]NH₂ displaying high μOR affinity were synthesized, in which Phe in either the third or fourth position was replaced by various derivatives of this amino acid (β³-Phe, homoPhe, β³-homoPhe and PhGly). The aim of this research was to examine the structural effects of such modifications on the bioactivity, and both experimental and theoretical methods were used. The binding of the cyclic analogues to all three OR types (μ, δ, κ) was assessed by radioligand competitive binding assay, and their functional activity was determined in a calcium mobilization assay. In order to provide structural hypotheses explaining the obtained experimental affinities, the complexes of the cyclic peptides with μOR were subjected to molecular modeling.     

新型四氢异喹啉类AChE抑制剂的虚拟筛选

通过考察GOLD对已知的TcAChE复合物中配体的结合构象的重现性,评价了其对AChE体系的分子对接的可靠性。运用分子设计方法,虚拟筛选了一系列具有不同碳链长度和取代基的四氢异喹啉类化合物(延胡索类生物碱corydaline开环衍生物),并分析了得分较高的配体分子与受体蛋白的相互结合作用。     

MS Binding Assays for Glycine Transporter 2 (GlyT2) Employing Org25543 as Reporter Ligand

This study describes the first binding assay for glycine transporter 2 (GlyT2) following the concept of MS Binding Assays. The selective GlyT2 inhibitor Org25543 was employed as a reporter ligand and it was quantified with a highly sensitive and rapid LC-ESI-MS/MS method. Binding of Org25543 at GlyT2 was characterized in kinetic and saturation experiments with an off-rate of 7.07×10⁻³ s⁻¹, an on-rate of 1.01×10⁶ M⁻¹ s⁻¹, and an equilibrium dissociation constant of 7.45 nM. Furthermore, the inhibitory constants of 19 GlyT ligands were determined in competition experiments. The validity of the GlyT2 affinities determined with the binding assay was examined by a comparison with published inhibitory potencies from various functional assays. With the capability for affinity determination towards GlyT2 the developed MS Binding Assays provide the first tool for affinity profiling of potential ligands and it represents a valuable new alternative to functional assays addressing GlyT2.