Therapeutic antibodies have been clinically applied to multiple of diseases including cancers, and their high efficacies are expected to accelerate the developments of various therapeutic antibodies in the future. However, in terms of physical properties including stability and affinity, “properties of excellent antibodies” are considered to be combinations of substantially complicated rules; therefore, efficient rules governing antibodies’ stability and affinity have not been clarified to date. Even if one succeeds in isolating candidate antibodies to be applicable as therapeutic antibodies, it is often necessary to modify amino acids to improve their stabilities and affinities. The optimization of antibody amino acid sequences in such candidate antibodies has been basically based on random or empirical modifications in large-scale mutagenesis screenings; but, this kind of effort-consuming strategy is not always the best and efficient way for the developments of optimal therapeutic antibodies. In other words, clarifying the rules of superior antibodies is an important theme in order to efficiently develop therapeutic antibodies that require higher stability and affinity.
Each B cell secretes its specific antibody with unique sequence due to somatic recombination of immunoglobulin genes, and such recombination results in substantially enormous diversity in antibody gene sequences; therefore, to obtain an overall picture of the antibody repertoires requires methodologies using next-generation sequencing. This research project is based on our original next-generation sequencing protocol for antibody repertoires derived from disease specimens as well as our original deep learning algorithm for the repertoire sequence data. This research project aims to establish a novel fundamental technology for the efficient developments of therapeutic antibodies. To this end, we will acquire as many antibody repertoire sequences as possible from lymphocytes, and analyze their amino acid sequences and properties by artificial intelligence technology. In addition, a group of researchers with significant experiences in evaluating physical properties of antibodies will measure physicochemical properties of candidate antibodies from various aspects. These physicochemical data will be re-used in the deep learning algorithm to achieve more accurate artificial intelligence technology.
The achievements of this results project are expected to be fundamental technologies that have large effects on a wide range of fields of therapeutic antibody developments. In the future, we will plan to apply this basic technology to identify and develop new anti-tumor antibodies, and to make clinically-proven therapeutic antibodies more stable and reliable. We aim to create basic technologies that can contribute to the new advanced biopharmaceuticals by conducting analysis focusing on malignant tumors which have bene major health problems worldwide.
Figure 1: Our Basic Technologies to be Combined in This Research Project.
Figure 2: Conceptual Diagram of This Research Project.
Figure 3: Members of This Research Project.