Speaker
Description
De-novo protein design allows a wider exploration of protein folding space than directed evolution and has been profitable employed in several fields of biochemistry and medicine, as metalloproteins design 1 and in-vitro diagnostics (IVD) 2. Among other design tasks, de novo protein binders can now be obtained with high success-rate that are highly specific towards a series of biological targets, thanks to the recent advances in the AI field [3,4,5]. Despite that, IVD industry still heavily relies on antibodies, as they are cheaper, and their production is well standardized. Therefore, a de-novo antigen may virtually speed-up and support the production of better antibodies. In this work, we present a workflow to scaffold small proteins (~50AA) around a known antigen epitope to elicit higher immune response in host organisms. The workflow is divided in two cyclic phases: generation and validation, each one using task-specific AI-models. The generation phase starts by creating several backbones of small scaffolds via diffusion models 3 around a chosen epitope, whose position and residues are held constant during the process. Subsequently, protein language models 4 are used on such backbones to assign the sequences. The structure of the generated sequences is then predicted through in-silico folding models 5 and the most promising sequences are screened in-silico for immunogenicity with MaSIF 6, a model evaluating protein-protein interaction sites based on geometric neural network. The methodology has been performed on one epitope of HPV-16, obtaining 6 different small proteins which are currently being screened in-vitro. The proposed workflow is fully generalizable to any epitope, and we envision that it can pose a novel methodology in the design of the de-novo antigens.
1 Chino, M., Di Costanzo, L.F., Leone, L. et al. “Designed Rubredoxin miniature in a fully artificial electron chain triggered by visible light”. Nat Commun 14, 2368 (2023). DOI: 10.1038/s41467-023-37941-8.
2 Chu, A.E., Lu, T. & Huang, PS. “Sparks of function by de novo protein design”. Nat Biotechnol 42, 203–215 (2024). DOI: 10.1038/s41587-024-02133-2.
3 Watson, J.L., Juergens, D., Bennett, N.R. et al. “De novo design of protein structure and function with RFdiffusion”. Nature 620, 1089–1100 (2023). DOI: 10.1038/s41586-023-06415-8.
4 Dauparas J. et al. “Robust deep learning–based protein sequence design using ProteinMPNN”. Science 378,49-56(2022). DOI: 10.1126/science.add2187.
5 Wu, R. et al. “High-resolution de novo structure prediction from primary sequence”. BioRxiv (2022). DOI: 10.1101/2022.07.21.500999.
6 Gainza, P., Sverrisson, F., Monti, F. et al. “Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning”. Nat Methods 17, 184–192 (2020). DOI: 10.1038/s41592-019-0666-6.
| Department | Dipartimento di Scienze Chimiche |
|---|
