Rotamer Profiling of Non-Canonical Amino Acids for Enhanced Ramachandran Mapping
Takanori AOKI *1, Ayumu MATSUDA2
1Research&Development Dept. Modeling&Informatics Group, PeptiDream Inc.
2Research&Development Dept. Medicinal Chemistry Group, PeptiDream Inc.
[Background]
The Ramachandran plot is a way to visualize energetically allowed regions for backbone dihedral angles (phi, psi angles). It is well known that natural amino acids show a certain trend of distribution in the Ramachandran plot. Expanding the interpretation to non-canonical amino acids, studies have used X-ray crystallography or NMR to elucidate these trends. Additionally, molecular dynamics simulations have been employed to generate conformations according to the Boltzmann distribution.
Peptide Discovery Platform System (PDPS) enables us to produce a highly diverse library of macrocyclic peptides (over 10 trillion) based on over 3,000 non-canonical amino acids. However, past studies on backbone conformation have not covered most of them. Therefore, we have been constructing a virtual Ramachandran plot database with molecular dynamics simulations for non-canonical amino acids available in PDPS. Moreover, statistical methods revealed the characteristics and diversity of their backbone conformation.
[Methods]
We performed REST2 simulations with Desmond (Schrodinger) for over 300 non-canonical amino acids as monomers containing various scaffolds or substitution groups. The distance between each monomer’s Ramachandran plots was defined by the Earth Mover’s Distance (EMD). Hierarchical clustering based on the calculated distances was performed to elucidate the backbone conformational diversity of the non-canonical amino acids. To share these insights with medicinal chemists, we visualized the data using a Spotfire dashboard.
[Results and Discussion]
The similarities among monomers identified through distance calculations and clustering were markedly distinct from those calculated with conventional chemical or physicochemical properties. Specifically, N-methylated alanine, proline, homo-proline and tetrahydro-isoquinoline-3-carboxylic acid (Tic) showed distinct characteristics. We will show and discuss these findings in more detail.
We applied this phi-psi database to design peptide backbone structure. We developed a backbone-design-system to rationally replace residues with rigid monomers based on phi-psi values of X-ray structures. In several cases, this approach led to the discovery of more potent peptides.
[Conclusions]
These insights have enabled us to optimize peptides for conformational rigidity and chemical stability. Moreover, we are exploring backbone scaffold hopping, including peptide mimetics.
In the poster presentation, we will show detailed Ramachandran plot data for specific non-canonical amino acids and the diversity of chemical space based on backbone differences.