Drug design utilizing 3D-RISM

Kota Murasaki (Mitsubishi Tanabe Pharma)

The Three-Dimensional Reference Interaction Site Model (3D-RISM) is a method for evaluating the three-dimensional distribution of solvent molecules using statistical mechanics. It is characterized by its ability to analyze solvation effects quickly and accurately. In MOE, 3D-RISM calculations can be easily performed via the GUI. In this presentation, we will introduce examples where 3D-RISM was applied to X-ray crystallography and Cryo-electron microscopy data obtained at Mitsubishi Tanabe Pharma Corporation. Additionally, we will present insights gained from analyzing PROTAC® ternary complexes using 3D-RISM.

Beyond Molecular Dynamics: Efficient Conformational Sampling in Nucleic Acids Using LowModeMD

Alain Ajamian (Chemical Computing Group)

Modeling nucleic acids, particularly RNA, presents challenges due to their dynamic nature, with biologically relevant conformations often differing significantly between unbound and bound states. This variability complicates predictions in silico, making efficient conformational sampling crucial. In this presentation, we investigate the efficacy of the LowModeMD method for enhancing conformational sampling in DNA and RNA systems compared to traditional molecular dynamics approaches.By focusing on low-frequency vibrational modes, LowModeMD enables rapid exploration of conformational space, facilitating the identification of biologically relevant structures. We demonstrate the utility of LowModeMD in analyzing bulged residues, sampling terminal and internal loops, cryptic pockets and large conformational changes caused by complexation.

In silico screening of Tcf21 agonists that can suppress progression of liver fibrosis

Noriaki Hirayama (Tokai University School of Medicine)

At present no medication is available that can directly suppress liver fibrosis in metabolic dysfunction-associated steatohepatitis. Since activated hepatic stellate cells (HSCs) lead to hepatic fibrogenesis, a deactivation factor is expected to suppress liver fibrosis. A transcription factor of Tcf21 has been recently identified as an effective deactivation factor. Tcf21 exerts its function by binding to both the response element of DNA and a conjugation factor such as Tcf3.

In order to undertake virtual screening forTcf21 agonists, the ternary complex of Tcf21/Tcf3/DNA was constructed by homology modeling. A detailed inspection of the model revealed a specific region of Tcf21 crucial for the function. A cluster of pharmacophores assigned to the region was used to screen a chemical database consisting of ca.7.7 million commercially available small-molecules. The binding affinities of the compounds to the Tcf3/DNA complex were evaluated by docking simulations by use of ASEDock. All calculations were performed using MOE.

In the activated HSCs, fibrogenic genes are expressed to produce collagens and other components of extracellular matrix leading to fibrosis. The gene coding -smooth muscle actin (Acta2) is one of such genes. The expression levels of Acta2 in HSCs treated with the screened compounds were measured to evaluate the deactivation activities. The results unequivocally showed that compounds with the high binding affinities to the Tcf3/DNA complex significantly decreased ACTA2 expression levels in HSCs. Research leading to clinical trials is now underway.

Medicinal chemistry on various inhibitors utilizing molecular modeling calculation with MOE

Takuya Kobayakawa (Laboratory for Biomaterials and Bioengineering Institute of Integrated Research, Institute of Science Tokyo)

Viral infections pose a significant threat to humanity, and the COVID-19 pandemic has reminded us of their severity. For chronic viral infections such as HIV infection and hepatitis B, there is an urgent need to develop therapeutic agents with novel mechanisms of action aimed at achieving a complete cure. Our laboratory has been conducting rational drug discovery research based on structural information of target proteins using the molecular modeling calculation software Molecular Operating Environment (MOE).

In this study, we performed in silico screening using MOE to search for novel inhibitors targeting HIV-1 capsid protein and HBV capsid protein. First, based on crystal structures of target proteins obtained from the Protein Data Bank, we selected candidate compounds considering both binding affinity and synthetic feasibility from an organic chemistry perspective.

As a result, we successfully identified MKN-1 and MKN-3 as HIV-1 capsid protein inhibitors, and Cpd4 and TKB-HBV-CA-001 as HBV capsid inhibitors. Furthermore, through structure-activity relationship studies using MOE docking simulations, we optimized these lead compounds and successfully created derivatives that showed superior antiviral activity compared to existing drugs. Notably, hybrid molecules of CD4 mimics and PEG also achieved improved pharmacokinetics.

In this presentation, I will introduce our comprehensive drug discovery research from the exploration to optimization of viral protein inhibitors using MOE, with specific examples.

1. Kobayakawa, T.; et al. Biomolecules 2021, 11, 208.
2. Kobayakawa, T.; et al. J. Med. Chem. 2021, 64, 1481-1496.
3. Kobayakawa, T.; et al. RSC Adv. 2023, 13, 2156-2167.
4. Kobayakawa, T.; et al. RSC Med. Chem. 2023, 14, 1973-1980.
5. Kobayakawa, T.; et al. Chem. Pharm. Bull. 2024, 72, 41-47.

Machine Learning Tools for mAbs Developability Predictions Utilizing MOE

Kozo Yoneda (Daiichi Sankyo, Inc.)

The developability evaluation for antibody-related modalities includes a wide range of factors, such as non-specific binding, self-interaction, hydrophobicity, electrostatic properties, and structural stability. This evaluation is crucial in biologics drug discovery research for the early identification of candidates with high manufacturability as well as favorable behavior in human, and it is providing numerous benefits like improvements in drug discovery speed and reductions in duration and costs associated with bioprocess development.

In particular, the in-silico approach to predict developability from sequence information has grown exponentially alongside advancements in antibody-related technology platforms and is actively researched in both academia and the pharmaceutical industry.

Recently, we established a wet evaluation system for high-throughput analysis of multiple developability parameters and created in-silico work flow for predicting developability by leveraging accumulated wet data and machine learning techniques. This machine learning utilizes protein feature quantities generated from software such as MOE. The details and accuracy of this in-silico developability work flow will be presented.

Comparative Analysis of Small Molecule Conformational Search Methods in Drug Discovery: Balancing Speed, Accuracy, and Interpretability

Jinxin Liu (Insilico Medicine)

Background: Molecular conformation plays a crucial role in determining the biological activity of small molecules in drug discovery. The accurate prediction of low-energy conformations and energy barriers is essential for understanding structure-activity relationships and optimizing drug candidates. With the rapid development of computational methods, various conformational search tools have emerged, ranging from traditional physics-based approaches to artificial intelligence-driven methods.

Methods: We conducted a comprehensive comparative study of multiple conformational search methodologies, including traditional approaches such as MOE and xtb, as well as AI-based methods like Auto3D. The evaluation focused on three key performance metrics: computational speed, accuracy of conformational prediction, and interpretability of results. Additionally, we assessed the precision of dihedral angle scanning capabilities across different platforms.

Results: Our analysis revealed that MOE demonstrates superior performance in achieving an optimal balance among search speed, accuracy, and interpretability compared to other evaluated methods. While AI-based approaches like Auto3D showed promising speed advantages, they often lacked the interpretability required for mechanistic understanding in drug design. Notably, in dihedral angle scanning applications, MOE's computational accuracy was found to be comparable to the gold-standard combination of xtb with Gaussian calculations.

Conclusions: MOE represents a well-balanced solution for small molecule conformational analysis in drug discovery workflows, offering researchers an optimal compromise between computational efficiency and chemical accuracy. The comparable performance to high-level quantum mechanical methods in dihedral scanning makes it particularly valuable for detailed conformational studies. These findings provide important guidance for selecting appropriate computational tools in structure-based drug design and optimization processes.

FMOe: A Tool for Streamlining Input Preparation and Analysis in FMO Calculations - Practical Applications

Kikuko Kamisaka (RIKEN Center for Integrative Medical Sciences)

The Fragment Molecular Orbital (FMO) method¹⁾ is a quantum chemical approach that divides large biomolecular systems into small fragments such as amino acid residues and ligands, and solves the electronic state of the entire molecule using a two-body correction approximation involving fragment monomers and dimers. Additionally, an inter-fragment interaction energy (IFIE) obtained from FMO calculation and its energy decomposition (Pair Interaction Energy Decomposition Analysis: PIEDA)²⁾ enable quantitative and precise evaluation of electrostatic interactions, hydrogen bonding, and hydrophobic effects. It is widely used for understanding molecular recognition and binding mechanisms. Software such as ABINIT-MP enables FMO calculations, and automatic fragmentation is supported for standard biomolecules like proteins. However, for complex systems such as covalent ligands, metal complexes, and medium-sized molecules containing nonstandard amino acids, the automatic fragmentation remains challenging and often requires specialized and labor-intensive manual setup.

To address this, we have developed FMOe (https://github.com/drugdesign/FMOe)³⁾, a graphical user interface (GUI) library that runs within the Molecular Operating Environment (MOE), to facilitate easier execution and analysis of FMO calculations. FMOe enables intuitive and flexible operations for fragment splitting and merging, and allows users to easily generate ABINIT-MP input files based on the specified fragment definitions. Furthermore, IFIE and PIEDA analyses can also be performed based on the three-dimensional structure of molecules. In parallel, our group has been continuously developing FMODB (https://drugdesign.riken.jp/FMODB/)⁴⁾ —a database for centrally managing FMO calculation results—since 2017 under the leadership of the FMO Drug Design Consortium (FMODD). The database has been widely adopted by the FMO community. In the latest update, we have added a new feature to FMOe that allows users to automatically download and load CPF files directly from FMODB using an FMODB ID, thereby streamlining the visualization of IFIE/PIEDA analysis results. Additional enhancements include an atomic charge analysis that evaluates fragment- and atom-level charge-based molecular coloring, as well as its list output functions; we have also implemented the automatic fragmentation for nucleic acid systems.

In this presentation, we will introduce the core functionalities of FMOe and highlight application cases involving covalent inhibitors and nucleic acids.

References:

1. Kitaura, K. et al., Chem. Phys. Lett., 1999, 313(3-4), 701-706.
2. Fedorov, D. G. et al., J. Comput. Chem., 2007, 28(1), 222-237.
3. Moriwaki, H. et al., J. Chem. Inf. Model. 2024, 64 (18), 6927-6937.
4. Takaya, D. et al., J. Chem. Inf. Model. 2021, 61(2), 777-794.

title

speaker

content