I will introduce an emerging field called Biopharmaceutical Informatics by providing different example case studies at various stages of discovery and development of biotherapeutics.

Understanding of bio-molecular interactions such as protein-ligand interactions are important for structure-based drug design. We have been focusing on not only molecular mechanics but also quantum mechanics (QM) that can deal with CH-π, π-π, and cation-π interactions. The Fragment Molecular Orbital (FMO) method can perform ab initio QM calculations for large-scale molecules by the fragmentation scheme. Generally, QM calculation takes highly computational cost for applying biological macromolecules such as protein-ligand complex structures. The FMO method enables us to efficiently perform ab initio quantum mechanical calculations by many-body expansion technique with fragmentation. The fragmentation technique of the FMO method leads to calculation of fragment-fragment interaction energies, which are useful for analyzing protein-ligand interactions. The pairwise fragment-fragment energies are called as inter-fragment interaction energy (IFIE), and its energy is decomposed by components of electrostatic, exchange repulsion, charge transfer, and dispersion interactions by pair interaction energy decomposition analysis (PIEDA). IFIE/PIEDA calculated by the FMO method are useful for analyzing the interactions. Currently, we are continually updating FMODB (https://drugdesign.riken.jp/FMODB/), which is a database for accessing IFIE/PIEDA of the bio-molecular interactions. The presentation will focus on features of FMODB and recent applications for the interaction analysis by the FMO method.

---

Dr. Daisuke Takaya works as a computational chemist in Graduate School of Pharmaceutical Science, Osaka University. He has been in charge of ligand-based and structure-based drug design for drug development projects. His work focuses on development of molecular modeling, machine learning, protein-ligand docking, database such as FMODB, and their applications for drug design. His latest interest is learning how to use PyTorch for constructing AI.

Marine natural products (MNPs) are promising pharmaceutical leads due to their structural diversity and potent biological activity. We have utilized chemical probes to investigate the target identification and mode of actions of MNPs. Aplyronine A (ApA) is an antitumor macrolide that stabilizes the protein–protein interaction between actin and tubulin to prevent spindle formation and mitosis in tumor cells. Negative staining with an electron microscope showed that the actin–ApA complex rapidly disrupted paclitaxel-stabilized microtubules. In addition, protein–protein docking and molecular dynamics simulations revealed that the actin–ApA complex preferred the helixes 11 and 12 as the binding position on tubulin. Further chemical biology studies on these MNPs will be presented.

---

Masaki Kita is currently a professor of Bioagricultural Sciences in Nagoya University since 2017. He received his PhD in 2005 from Nagoya University, and has investigated various natural products that regulate biologically and physiologically intriguing phenomena. He was an assistant professor at Nagoya University (2001–2007) and University of Tsukuba (2007–2011), and was promoted to an associate professor in 2011. His current research interests focus on the target molecules and mode of actions of natural products by using their chemical probes.

mAb candidates identified from high-throughput screening or binding affinity optimization often present liabilities for developability, such as aggregation-prone regions or poor solution behavior. In this work, we optimized an integrin α11 binding mAb for developability using homology modeling and rational design where reducing hydrophobic surface patches improved HIC behavior. A retrospective data analysis demonstrates that 3D descriptors and multi-parameter models can screen candidates and enrich libraries with favorable developability properties for a range of biotherapeutics.

---

Andrew Henry studied with Professor Tony Rees at the University of Bath, using homology modeling to help reduce the immunogenicity of therapeutic antibodies. This work continued at Oxford Molecular and Millennium Pharmaceuticals. Andrew has worked for Chemical Computing Group customer support group since 2003 and is now a Principal Scientist, (focusing on the biologics field).

RNA viruses include genetic variation to ensure their survival, which is attributed to high mutation rates, high yields, and short replication times. The genetic variation often results in alterations in immune sensitivity, drug sensitivity, cell tropism, and host range, causing uncontrollability of the pathogen and emerging/re-emerging infections. To better understand the changes of the viruses and develop effective methods to control the moving targets, it is essential to obtain information on the changes in viral protein structure. We will present the molecular basis of the immune sensitivity, cell tropism, or host range of the RNA virus, human immunodeficiency virus type 1 or influenza A virus, based on structure analysis by computational science.

---

Masaru Yokoyama has received a Ph.D. in Engineering from Keio University in 1997. From 2002 to 2004, he studied the crystallization and molecular modeling of an HIV-1 reverse transcriptase as a research resident (Japanese Foundation for AIDS Prevention) at the National Institute of Infectious Diseases. From 2004 to 2005, he studied the allosteric regulation for nucleoside analogue resistance and molecular modeling of an HIV-1 reverse transcriptase as a postdoctoral fellow at the Institute of Tropical Medicine, Nagasaki University. Since 2005 at the National Institute of Infectious Diseases, he has been studying computational virology of RNA viruses.

Protein-protein binding affinity prediction is key in the practice of antibody affinity maturation. Molecular dynamics (MD) coupled with implicit solvent models are often used in affinity evaluation for antibody design upon mutations near the hotspots. This presentation will focus on the implementation of Hamiltonian replica exchange (H-REMD) and illustrate how enhanced sampling serves as an effective tool in understanding the antibody-antigen complex.

---

Dr. Qian obtained her PhD degree at Yale University in Jorgensen lab. She later joined the Broad Institute of MIT and Harvard working on multiple drug discovery projects. She is currently leading the computational chemistry group at VivaBiotech, Shanghai.

Next-generation multi-specific antibody therapeutics have emerged as successful treatment options for complex, multifactorial illnesses. As with mono-specific antibodies, optimization of pharmacokinetic properties is also of great importance for multi-specific biologics. By profiling the surface patch landscape of the multi-specific cross-over dual variable Ig-like (CODV-Ig) format, we were able to correlate in silico derived molecular properties with in vivo clearance, establishing the principles of pharmacokinetic predictions for multi-specific antibodies.

Antibodies are indisputably the most successful reagents in molecular-targeting therapy. However, the use of antibodies has been limited due to the biophysical properties and the cost to manufacture. To enable new applications where antibodies show some limitations, we have developed an alternative-binding molecule with non-immunoglobulin domain. The molecule is a helix-loop-helix (HLH) peptide, which is stable against enzyme degradations in vivo and is too small to show immunogenicity. Here, we introduce our HLH molecular-targeting peptides that show antibody-like functions, high affinity and high specificity for the targeted proteins.

Since the HLH peptide folds by virtue of hydrophobic and electrostatic interactions between the amino acid residues positioned inside the molecule, the outside solvent-exposed residues are possible to be mutated with a variety of amino acids to give a combinatorial library of the HLH peptides. Based on our technology of phage-displayed libraries for antibodies, we constructed a phage-displayed library of the HLH peptides. The library was screened against G-CSF receptor to give a binding peptide, which was cyclized by a thioether linkage between the N- and C-termini. The cyclic peptide showed a strong binding affinity (Kd of 4 nM) to the receptor and a long half-life (>2 weeks) in mouse sera, proving an enzyme-resistant property. Immunization of the HLH peptide to mice showed no induction of the antibody production (non-immunogenic). We have applied our HLH peptide libraries for CTLA4, VEGF, IgG/Fc, and kinases to obtain their molecular-targeting HLH peptides. In addition, we used the HLH peptide as a scaffold for generating cell permeable targeting peptides through bi-functional grafting: epitope grafting to provide binding activity and arginine grafting to endow cell-permeability. The HLH peptides provide insights into de novo peptide-based drug discovery and then would be a new therapeutic modality.

---

Prof. Ikuo Fujii earned his Ph.D. on pharmaceutical science at Kyushu University in 1986, and then he was an assistant professor of the University. He was a postdoctoral associate at the Rockefeller University from 1988-1989. He was moved to The Scripps Research Institute in 1989. In 1991, he was appointed to a senior research scientist to direct the antibody-engineering group in Protein Engineering Research Institute (Osaka). He was moved to Biomolecuar Engineering Research Institute (Osaka) and promoted to the research director in Department of Bioorganic Chemistry in 1996. In 2003, he was appointed to a professor in Osaka Prefecture University (later renamed Osaka Metropolitan University).