Conformational Selectivity of ITK Inhibitors: Insights from Molecular Dynamics Simulations.

Interleukin-2-inducible T-cell kinase (ITK) regulates the response to T-cell receptor signaling and is a drug target for inflammatory and immunological diseases. Molecules that bind preferentially to the active form of ITK have low selectivity between kinases, whereas those that bind preferentially to the inactive form have high selectivity for ITK. Therefore, computational methods to predict the conformational selectivity of compounds are required to design highly selective ITK inhibitors. In this study, we performed absolute binding free-energy perturbation (ABFEP) simulations for 11 compounds on both active and inactive forms of ITK, and the calculated binding free energies were compared with experimental data. The conformational selectivity of 10 of the 11 compounds was correctly predicted using ABFEP. To investigate the mechanism underlying the stabilization of the active and inactive structures by the compounds, we performed extensive, conventional molecular dynamics simulations, which revealed that the compound-induced stabilization of the P-loop and linkage of conformational changes in L489, V419, F501, and M410 upon compound binding were critical factors. A guideline for designing inactive-form binders is proposed based on these key structural factors. The ABFEP and the created guidelines are expected to facilitate the discovery of highly selective ITK inhibitors.

Selective Inhibitor Design for Kinase Homologs Using Multiobjective Monte Carlo Tree Search.

Designing highly selective molecules for a drug target protein is a challenging task in drug discovery. This task can be regarded as a multiobjective problem that simultaneously satisfies criteria for various objectives, such as selectivity for a target protein, pharmacokinetic endpoints, and drug-like indices. Recent breakthroughs in artificial intelligence have accelerated the development of molecular structure generation methods, and various researchers have applied them to computational drug designs and successfully proposed promising drug candidates. However, designing efficient selective inhibitors with releasing activities against various homologs of a target protein remains a difficult issue. In this study, we developed a de novo structure generator based on reinforcement learning that is capable of simultaneously optimizing multiobjective problems. Our structure generator successfully proposed selective inhibitors for tyrosine kinases while optimizing 18 objectives consisting of inhibitory activities against 9 tyrosine kinases, 3 pharmacokinetics endpoints, and 6 other important properties. These results show that our structure generator and optimization strategy for selective inhibitors will contribute to the further development of practical structure generators for drug designs.

gr Predictor: A Deep Learning Model for Predicting the Hydration Structures around Proteins.

Among the factors affecting biological processes such as protein folding and ligand binding, hydration, which is represented by a three-dimensional water site distribution function around the protein, is crucial. The typical methods for computing the distribution functions, including molecular dynamics simulations and the three-dimensional reference interaction site model (3D-RISM) theory, require a long computation time ranging from hours to tens of hours. Here, we propose a deep learning (DL) model that rapidly estimates the distribution functions around proteins obtained using the 3D-RISM theory from the protein 3D structure. The distribution functions predicted using our DL model are in good agreement with those obtained using the 3D-RISM theory. Particularly, the coefficient of determination between the distribution function obtained by the DL model and that obtained using the 3D-RISM theory is approximately 0.98. Furthermore, using a graphics processing unit, the prediction by the DL model is completed in less than 1 min, more than 2 orders of magnitude faster than the calculation time of the 3D-RISM theory. The position of water molecules around the protein was estimated based on the distribution function obtained by our DL model, and the position of waters estimated by our DL model was in good agreement with that of water molecules estimated using the 3D-RISM theory and of crystallographic waters. Therefore, our DL model provides a practical and efficient way to calculate the three-dimensional water site distribution functions and to estimate the position of water molecules around the protein. The program called "gr Predictor" is available under the GNU General Public License from https://github.com/YoshidomeGroup-Hydration/gr-predictor.

Effect of Water Molecules on the Activating S810L Mutation of the Mineralocorticoid Receptor.

The mineralocorticoid receptor (MR) is a nuclear receptor whose endogenous ligands are mineralocorticoids, a type of steroid hormone. The activating S810L mutation is known to cause severe early-onset and pregnancy-related hypertension. Progesterone binds to the wild-type (WT) MR as a passive antagonist with fast dissociation; however, it binds to the S810L mutant as a full agonist with slow dissociation. The switch in the biological activity of progesterone is considered to be one of the causes of the disease. First, we used steered molecular dynamics simulations to analyze the dissociation process of progesterone for the WT and the S810L mutant. Progesterone in the WT dissociated from the ligand-binding pocket with a weak force in comparison with progesterone in the S810L mutant due to the large inflow of water molecules into the pocket. Therefore, we used conventional molecular dynamics simulations for the ligand-free structures of the WT and the S810L mutant to investigate the effect of the mutation on the inflow of water. In the WT, water molecules enter the ligand-binding pocket in two ways: in the vicinity of (i) Arg817 and (ii) Ser810. In contrast, few water molecules enter the pocket in the S810L mutant because of the large size and hydrophobic nature of the Leu810 side chain. Fast dissociation is a common feature among passive antagonists of MR; therefore, we inferred that the water inflow could be responsible for the dissociation kinetics of progesterone in the WT and the S810L mutant.

Structure-based screening combined with computational and biochemical analyses identified the inhibitor targeting the binding of DNA Ligase 1 to UHRF1.

The accumulation of epigenetic alterations is one of the major causes of tumorigenesis. Aberrant DNA methylation patterns cause genome instability and silencing of tumor suppressor genes in various types of tumors. Therefore, drugs that target DNA methylation-regulating factors have great potential for cancer therapy. Ubiquitin-like containing PHD and RING finger domain 1 (UHRF1) is an essential factor for DNA methylation maintenance. UHRF1 is overexpressed in various cancer cells and down-regulation of UHRF1 in these cells reactivates the expression of tumor suppressor genes, thus UHRF1 is a promising target for cancer therapy. We have previously shown that interaction between the tandem Tudor domain (TTD) of UHRF1 and DNA ligase 1 (LIG1) di/trimethylated on Lys126 plays a key role in the recruitment of UHRF1 to replication sites and replication-coupled DNA methylation maintenance. An arginine binding cavity (Arg-binding cavity) of the TTD is essential for LIG1 interaction, thus the development of inhibitors that target the Arg-binding cavity could potentially repress UHRF1 function in cancer cells. To develop such an inhibitor, we performed in silico screening using not only static but also dynamic metrics based on all-atom molecular dynamics simulations, resulting in efficient identification of 5-amino-2,4-dimethylpyridine (5A-DMP) as a novel TTD-binding compound. Crystal structure of the TTD in complex with 5A-DMP revealed that the compound stably bound to the Arg-binding cavity of the TTD. Furthermore, 5A-DMP inhibits the full-length UHRF1:LIG1 interaction in Xenopus egg extracts. Our study uncovers a UHRF1 inhibitor which can be the basis of future experiments for cancer therapy.

Comprehensive 3D-RISM analysis of the hydration of small molecule binding sites in ligand-free protein structures.

Hydration is a critical factor in the ligand binding process. Herein, to examine the hydration states of ligand binding sites, the three-dimensional distribution function for the water oxygen site, gO (r), is computed for 3,706 ligand-free protein structures based on the corresponding small molecule-protein complexes using the 3D-RISM theory. For crystallographic waters (CWs) close to the ligand, gO (r) reveals that several CWs are stabilized by interaction networks formed between the ligand, CW, and protein. Based on the gO (r) for the crystallographic binding pose of the ligand, hydrogen bond interactions are dominant in the highly hydrated regions while weak interactions such as CH-O are dominant in the moderately hydrated regions. The polar heteroatoms of the ligand occupy the highly hydrated and moderately hydrated regions in the crystallographic (correct) and wrongly docked (incorrect) poses, respectively. Thus, the gO (r) of polar heteroatoms may be used to distinguish the correct binding poses.

High-Precision Atomic Charge Prediction for Protein Systems Using Fragment Molecular Orbital Calculation and Machine Learning.

Here, we have constructed neural network-based models that predict atomic partial charges with high accuracy at low computational cost. The models were trained using high-quality data acquired from quantum mechanics calculations using the fragment molecular orbital method. We have succeeded in obtaining highly accurate atomic partial charges for three representative molecular systems of proteins, including one large biomolecule (approx. 2000 atoms). The novelty of our approach is the ability to take into account the electronic polarization in the system, which is a system-dependent phenomenon, being important in the field of drug design. Our high-precision models are useful for the prediction of atomic partial charges and expected to be widely applicable in structure-based drug designs such as structural optimization, high-speed and high-precision docking, and molecular dynamics calculations.

Monte Carlo simulations on atropisomerism of thienotriazolodiazepines applicable to slow transition phenomena using potential energy surfaces by ab initio molecular orbital calculations.

Compounds with a medium-sized flexible ring often show atropisomerism that is caused by the high-energy barriers between long-lived conformers that can be isolated and often have different biological properties to each other. In this study, the frequency of the transition between the two stable conformers, aS and aR, of thienotriazolodiazepine compounds with flexible 7-membered rings was estimated computationally by Monte Carlo (MC) simulations and validated experimentally by NMR experiments. To estimate the energy barriers for transitions as precisely as possible, the potential energy (PE) surfaces used in the MC simulations were calculated by molecular orbital (MO) methods. To accomplish the MC simulations with the MO-based PE surfaces in a practical central processing unit (CPU) time, the MO-based PE of each conformer was pre-calculated and stored before the MC simulations, and then only referred to during the MC simulations. The activation energies for transitions calculated by the MC simulations agreed well with the experimental ΔG determined by the NMR experiments. The analysis of the transition trajectories of the MC simulations revealed that the transition occurred not only through the transition states, but also through many different transition paths. Our computational methods gave us quantitative estimates of atropisomerism of the thienotriazolodiazepine compounds in a practical period of time, and the method could be applicable for other slow-dynamics phenomena that cannot be investigated by other atomistic simulations.

Synthesis and Biological Evaluations of Novel Human Parathyroid Hormone 1 Receptor (hPTHR1) Agonists Bearing Bicyclic Aromatic Moiety.

We have previously shown that the small molecule hPTHR1 agonist PCO371 (1) orally and dose-dependently induces PTH-like calcemic and hypophostemic activity in thyroparathyroidectomized rats. Compound 2a, bearing a bicyclic aromatic ring, was identified as a novel hPTHR1 agonist during hit to lead modification. It showed moderate PTHR1 agonistic activity with an EC20 value of 15 µM, and its metabolic stability in human liver microsome (hLM) as well as its solubility in phosphate buffer (PPb) and Fasted state simulated intestinal fluid (FaSSIF) were found to be poor. As results of the initial derivatization of 2a, we identified the indole derivatives as another scaffold. In this article, we report on the structure-activity relationship (SAR), structure-metabolism relationship (SMR), and structure-solubility relationship (SSR) of bicyclic aromatic derivatives, and the in vivo efficacy of 2j.

Effects of fluorines on nonsecosteroidal vitamin D receptor agonists.

From our research of nonsecosteroidal vitamin D(3) derivatives with gamma hydroxy carboxylic acid, we identified compound 6, with two CF(3) groups in the side chain, as a most potent vitamin D receptor (VDR) agonist that shows superagonistic activity in VDRE reporter gene assay, MG-63 osteocalcin production assay and HL-60 cell differentiation assay. Compound 6 demonstrated that fluorination is as effective in the case of our nonsecosteroidal scaffold as in the case of secosteroidal VD(3) analogs. X-ray analysis of the VDR with compound 6 revealed all of the six fluorine atoms of the hexafluoropropanol (HFP) moiety in the side chain effectively interacting with the VDR by both steric (van der Waals) and electrostatic (hydrogen bond, NH-F and CH-F) interactions. The HFP moiety of 6 effectively interacts with helix 12 (H12) of the VDR and stabilizes the position and the orientation of H12, which could result in stabilizing the coactivator and enhancing the VDR agonistic activity.

A series of nonsecosteroidal vitamin D receptor agonists for osteoporosis therapy.

In an extension of our study on gamma hydroxy carboxylic acid analogs, we explored a series of nonsecosteroidal vitamin D receptor (VDR) agonists in which 1,3-diol of 1,25(OH)2D3 had been replaced by aryl acetic acid. These analogs showed very potent activity in vitro compared with 1,25(OH)2D3. An X-ray analysis of 8d showed that the inserted phenyl ring well mimicked the folded methylene linker of the gamma hydroxy carboxylic acid moiety but the carboxylic acid of 8d interacted with VDR in a different manner from gamma hydroxy carboxylic acids. Through our in vivo screening in an osteoporosis rat model using immature rats, we identified a potent active vitamin D3 analog, compound 7e. In mature rats of the same model, compound 7e also showed good PK profiling and excellent ability to prevent bone mineral density loss without severe hypercalcemia. Our nonsecosteroidal VDR agonist 7e (CH5036249) could be a possible new drug candidate for treating osteoporosis in human.

Structure-Based Affinity Maturation of Antibody Based on Double-Point Mutations.

Structure-based site-directed affinity maturation of antibodies can be expanded by multiple-point mutations to obtain various mutants. However, selecting the appropriate number of promising mutants for experimental evaluation from the vast number of combinations of multiple-point mutations is challenging. In this report, we describe how to narrow candidate mutants using the so-called weak interaction analysis such as CH-π and CH-O in addition to widely recognized interactions such as hydrogen bonds.

AI-driven molecular generation of not-patented pharmaceutical compounds using world open patent data.

Developing compounds with novel structures is important for the production of new drugs. From an intellectual perspective, confirming the patent status of newly developed compounds is essential, particularly for pharmaceutical companies. The generation of a large number of compounds has been made possible because of the recent advances in artificial intelligence (AI). However, confirming the patent status of these generated molecules has been a challenge because there are no free and easy-to-use tools that can be used to determine the novelty of the generated compounds in terms of patents in a timely manner; additionally, there are no appropriate reference databases for pharmaceutical patents in the world. In this study, two public databases, SureChEMBL and Google Patents Public Datasets, were used to create a reference database of drug-related patented compounds using international patent classification. An exact structure search system was constructed using InChIKey and a relational database system to rapidly search for compounds in the reference database. Because drug-related patented compounds are a good source for generative AI to learn useful chemical structures, they were used as the training data. Furthermore, molecule generation was successfully directed by increasing and decreasing the number of generated patented compounds through incorporation of patent status (i.e., patented or not) into learning. The use of patent status enabled generation of novel molecules with high drug-likeness. The generation using generative AI with patent information would help efficiently propose novel compounds in terms of pharmaceutical patents. Scientific contribution: In this study, a new molecule-generation method that takes into account the patent status of molecules, which has rarely been considered but is an important feature in drug discovery, was developed. The method enables the generation of novel molecules based on pharmaceutical patents with high drug-likeness and will help in the efficient development of effective drug compounds.

Systematic SAR study of the side chain of nonsecosteroidal vitamin D3 analogs.

A series of nonsecosteroidal vitamin D(3) analogs with carboxylic acid were explored. Through our systematic SAR studies on the side chain moiety, compound 6b was identified as the optimal compound showing excellent vitamin D receptor (VDR) agonistic activity. Compound 6b had the diethyl group in the terminal which was bound by (E)-olefin linker to the bisphenyl core. Calculating the volume of the side chain showed that the diethyl group in 6b filled the hydrophobic region of VDR with the ideal packing coefficient based on the 55% rule, and that this resulted in the most potent in vitro activity.

Machine learning to estimate the local quality of protein crystal structures.

Low-resolution electron density maps can pose a major obstacle in the determination and use of protein structures. Herein, we describe a novel method, called quality assessment based on an electron density map (QAEmap), which evaluates local protein structures determined by X-ray crystallography and could be applied to correct structural errors using low-resolution maps. QAEmap uses a three-dimensional deep convolutional neural network with electron density maps and their corresponding coordinates as input and predicts the correlation between the local structure and putative high-resolution experimental electron density map. This correlation could be used as a metric to modify the structure. Further, we propose that this method may be applied to evaluate ligand binding, which can be difficult to determine at low resolution.

Structure-activity relationships of bioisosteric replacement of the carboxylic acid in novel androgen receptor pure antagonists.

A series of 5,5-dimethylthiohydantoin derivatives were synthesized and evaluated for androgen receptor pure antagonistic activities for the treatment of hormone refractory prostate cancer. CH4933468 (32d) with a sulfonamide side chain not only exhibited antagonistic activity with no agonistic activity in the reporter gene assay but also inhibited the growth of bicalutamide-resistant cell lines. This compound also inhibited tumor growth of the LNCaP xenograft in mice dose-dependently.

Design and synthesis of an androgen receptor pure antagonist (CH5137291) for the treatment of castration-resistant prostate cancer.

A series of 5,5-dimethylthiohydantoin derivatives were synthesized and evaluated for androgen receptor pure antagonistic activities for the treatment of castration-resistant prostate cancer. Since CH4933468, which we reported previously, had a problem with agonist metabolites, novel thiohydantoin derivatives were identified by applying two strategies. One was the replacement of the alkylsulfonamide moiety by a phenylsulfonamide to avoid the production of agonist metabolites. The other was the replacement of the phenyl ring with a pyridine ring to improve in vivo potency and reduce hERG affinity. Pharmacological assays indicated that CH5137291 (17b) was a potent AR pure antagonist which did not produce the agonist metabolite. Moreover, CH5137291 completely inhibited in vivo tumor growth of LNCaP-BC2, a castration-resistant prostate cancer model.

Design and synthesis of peptidomimetic factor VIIa inhibitors.

Selective factor VIIa-tissue factor complex (FVIIa/TF) inhibition is regarded as a promising target for developing new anticoagulant drugs. In previous reports, we described a S3 subsite found in the X-ray crystal structure of compound 2 that bound to FVIIa/soluble tissue factor (sTF). Based on the X-ray crystal structure information and with the aim of improving the inhibition activity for FVIIa/TF and selectivity against other serine proteases, we synthesized derivatives by introducing substituents at position 5 of the indole ring of compound 2. Among them, compound 16 showed high selectivity against other serine proteases. Contrary to our expectations, compound 16 did not occupy the S3-subsite; X-ray structure analysis revealed that compound 16 improved selectivity by forming hydrogen bonds with Gln217, Thr99 and Asn100.

Structure-based design and discovery of novel anti-tissue factor antibodies with cooperative double-point mutations, using interaction analysis.

The generation of a wide range of candidate antibodies is important for the successful development of drugs that simultaneously satisfy multiple requirements. To find cooperative mutations and increase the diversity of mutants, an in silico double-point mutation approach, in which 3D models of all possible double-point mutant/antigen complexes are constructed and evaluated using interaction analysis, was developed. Starting from an antibody with very high affinity, four double-point mutants were designed in silico. Two of the double-point mutants exhibited improved affinity or affinity comparable to that of the starting antibody. The successful identification of two active double-point mutants showed that a cooperative mutation could be found by utilizing information regarding the interactions. The individual single-point mutants of the two active double-point mutants showed decreased affinity or no expression. These results suggested that the two active double-point mutants cannot be obtained through the usual approach i.e. a combination of improved single-point mutants. In addition, a triple-point mutant, which combines the distantly located active double-point mutation and an active single-point mutation collaterally obtained in the process of the double-point mutation strategy, was designed. The triple-point mutant showed improved affinity. This finding suggested that the effects of distantly located mutations are independent and additive. The double-point mutation approach using the interaction analysis of 3D structures expands the design repertoire for mutants, and hopefully paves a way for the identification of cooperative multiple-point mutations.

kGCN: a graph-based deep learning framework for chemical structures.

Deep learning is developing as an important technology to perform various tasks in cheminformatics. In particular, graph convolutional neural networks (GCNs) have been reported to perform well in many types of prediction tasks related to molecules. Although GCN exhibits considerable potential in various applications, appropriate utilization of this resource for obtaining reasonable and reliable prediction results requires thorough understanding of GCN and programming. To leverage the power of GCN to benefit various users from chemists to cheminformaticians, an open-source GCN tool, kGCN, is introduced. To support the users with various levels of programming skills, kGCN includes three interfaces: a graphical user interface (GUI) employing KNIME for users with limited programming skills such as chemists, as well as command-line and Python library interfaces for users with advanced programming skills such as cheminformaticians. To support the three steps required for building a prediction model, i.e., pre-processing, model tuning, and interpretation of results, kGCN includes functions of typical pre-processing, Bayesian optimization for automatic model tuning, and visualization of the atomic contribution to prediction for interpretation of results. kGCN supports three types of approaches, single-task, multi-task, and multi-modal predictions. The prediction of compound-protein interaction for four matrixmetalloproteases, MMP-3, -9, -12 and -13, in the inhibition assays is performed as a representative case study using kGCN. Additionally, kGCN provides the visualization of atomic contributions to the prediction. Such visualization is useful for the validation of the prediction models and the design of molecules based on the prediction model, realizing "explainable AI" for understanding the factors affecting AI prediction. kGCN is available at https://github.com/clinfo.