We are elucidating the laws of properties, features, and functions of protein to rationally design biomolecules, especially de novo proteins.
Proteins are polymers composed of 20 different amino acids, and even a protein composed of only 80 amino acids has more than 10 to the 100th power of possible sequences (i.e. the number of atoms in the entire universe is 10 to the 80th power). While the high diversity of proteins allows them to exhibit a wide variety of functions, it is difficult to rationally design de novo proteins without a thorough understanding of their properties and features that support their functions. To this end, we aim to elucidate the laws related to protein by combining large-scale analytical methods with machine learning, including deep learning.
Talk about Tsuboyama Nature 2023 on Machine learning for protein engineering seminar
Protein-protein interactions are the most fundamental and important functions of proteins. Protein-protein interactions are relevant to all biological functions such as signal transduction and metabolism, and are also important for designing functional artificial proteins (e.g., de novo proteins that can prevent new coronavirus infections or de novo proteins that act as switches, etc.). Thus, we will elucidate what features are important for protein-protein interactions.
Proteins are constantly being translated (synthesized) and degraded by autophagy and/or proteasome system to keep protein quality. We humans have about 20,000 genes, of which about 600-800 are involved in protein degradation.
While predictive models of protein folding structures (in preparation) have been constructed using data from Tsuboyama et al 2022, it remains unclear what proteins are more stable or unstable in vivo (in cells). We aim to elucidate the mechanism of protein lifespan.
One of the major clinical advantages of de novo proteins over small molecules is the ability to control protein-protein interactions. For example, in cancer, it is expected to control interactions such as Ras-Raf or Myc-Max. On the other hand, most of such important protein-protein interactions occur in the cytoplasm, not on the plasma membrane or outside the cell. Therefore, to control such important protein-protein interactions, it is necessary to efficiently transfer de novo proteins into the cytoplasm. For the future medical applications of engineered proteins, we are exploring the features necessary and sufficient for such cytosolic delivery of proteins