This article introduces RFpeptides, a novel deep learning pipeline for the de novo design of high-affinity protein-binding macrocycles. The current methods for creating macrocyclic binders…
This article introduces RFpeptides, a novel deep learning pipeline for the de novo design of high-affinity protein-binding macrocycles. The current methods for creating macrocyclic binders rely heavily on resource-intensive screening processes that offer limited control over binding modes.
RFpeptides utilizes a denoising diffusion-based approach to overcome these limitations. The researchers tested their approach on four diverse proteins, designing and testing a small number of macrocycles for each. The results of the RFpeptides pipeline were promising, with the designed macrocycles demonstrating medium to high affinity for all target proteins.
Notably, the team successfully designed a high-affinity binder (Kd < 10 nM) for Rhombotarget A (RbtA), even when starting from the predicted target structure. The study further validated the accuracy of the computational models by comparing them to X-ray structures of macrocycle-bound complexes.
The X-ray structures of the macrocycle complexes with myeloid cell leukemia 1, γ-aminobutyric acid type A receptor-associated protein, and RbtA closely matched the computational models, with a Cα root-mean-square deviation of less than 1.5 Å. The development of macrocyclic peptides holds significant potential in therapeutics, bridging the gap between small-molecule drugs and larger biologics.
Macrocycles can access intracellular targets while also effectively targeting proteins that lack deep hydrophobic pockets. The RFpeptides pipeline offers a significant advancement by providing a framework for the rapid, custom design of macrocyclic peptides. This has the potential to be applied in various diagnostic and therapeutic applications.
In conclusion, RFpeptides represents a significant step forward in the field of protein design. By employing a deep learning approach, the research team was able to overcome limitations in existing methods and successfully design high-affinity macrocyclic binders. The accuracy and efficiency of the pipeline, as demonstrated by the experimental validation, highlight its potential for broad applications in diagnostics and therapeutics.