AlphaFold accurately predicts distinct conformations based on the oligomeric state of a de novo designed protein

Abstract

Using the molecular modeling program Rosetta, we designed a de novo protein, called SEWN0.1, which binds the heterotrimeric G protein Gα q. The design is helical, well-folded, and primarily monomeric in solution at a concentration of 10 μM. However, when we solved the crystal structure of SEWN0.1 at 1.9 Å, we observed a dimer in a conformation incompatible with binding Gα q . Unintentionally, we had designed a protein that adopts alternate conformations depending on its oligomeric state. Recently, there has been tremendous progress in the field of protein structure prediction as new methods in artificial intelligence have been used to predict structures with high accuracy. We were curious if the structure prediction method AlphaFold could predict the structure of SEWN0.1 and if the prediction depended on oligomeric state. When AlphaFold was used to predict the structure of monomeric SEWN0.1, it produced a model that resembles the Rosetta design model and is compatible with binding Gα q , but when used to predict the structure of a dimer, it predicted a conformation that closely resembles the SEWN0.1 crystal structure. AlphaFold's ability to predict multiple conformations for a single protein sequence should be useful for engineering protein switches.

De Novo Design of Polyhedral Protein Assemblies: Before and After the AI Revolution

“…DL-based structure prediction tools can be helpful in selecting the test set for experimental validation, and it is now common to see that structure prediction tools are often integrated into the design pipeline. [66,91,92] Neural network tools for structural prediction provide independent confirmation that the newly designed sequences have a reasonable chance of folding correctly and forming the desired contacts. The first high-quality predictions of single protein chains were achieved by AlphaFold2, reaching near-atomic accuracy of the protein backbone.…”

“…To further validate the design methods and to evaluate their success at particular design tasks, selected designs that satisfy user‐defined criteria must be experimentally tested. DL‐based structure prediction tools can be helpful in selecting the test set for experimental validation, and it is now common to see that structure prediction tools are often integrated into the design pipeline [66,91,92] . Neural network tools for structural prediction provide independent confirmation that the newly designed sequences have a reasonable chance of folding correctly and forming the desired contacts.…”

Engineering strategies and challenges of endolysin as an antibacterial agent against Gram‐negative bacteria

“…Following optimization by design-test-learn loop, their hit rate improved by 10-fold (Lipsh-Sokolik et al, 2023). Cummins et al (2023) designed stable helical proteins using a de novo design method that effectively inhibits oncogenic G proteins. They confirmed that computational protein design, in combination with motif grafting, can be used to directly generate functional proteins without further optimization via high throughput screening or selection.…”

A journey from molecule to physiology and in silico tools for drug discovery targeting the transient receptor potential vanilloid type 1 (TRPV1) channel

“…These conformations are posited to bear resemblance to native structures, acknowledging that our current methodologies, including cryo-EM and crystallography, may not provide a definitive picture of native states, as alluded to by recent discussions in the field ( Shoemaker and Ando, 2018 ). In addition, it has been tested that the ability of AF2 to predict multiple conformations for a given protein sequence holds potential for engineering protein and mutant induced activity ( Cummins et al, 2022 ). Another success derived from AF2 is the recent update of the Membranome Database 3.0, which incorporates models generated by AF2 and validated with experimental information ( Lomize et al, 2022 ).…”