Small-molecule binding and sensing with a designed protein family

Abstract

Despite transformative advances in protein design with deep learning, the design of small-molecule–binding proteins and sensors for arbitrary ligands remains a grand challenge. Here we combine deep learning and physics-based methods to generate a family of proteins with diverse and designable pocket geometries, which we employ to computationally design binders for six chemically and structurally distinct small-molecule targets. Biophysical characterization of the designed binders revealed nanomolar to low micromolar binding affinities and atomic-level design accuracy. The bound ligands are exposed at one edge of the binding pocket, enabling thede novodesign of chemically induced dimerization (CID) systems; we take advantage of this to create a biosensor with nanomolar sensitivity for cortisol. Our approach provides a general method to design proteins that bind and sense small molecules for a wide range of analytical, environmental, and biomedical applications.

Rationalizing diverse binding mechanisms to the same protein fold: insights for ligand recognition and biosensor design

“…Indeed, aligning the known WIN conformer to its known Hbond acceptor geometry, followed by Rosetta FastDesign 42 , is sufficient to generate designs that recognize WIN with a nanomolar limit of detection. Developing new deep learning methods to learn and design small molecule-protein interactions is in vogue 15,[43][44][45][46][47][48] . We suggest that comparable attention should be placed with the choice of ligand conformer and rigid body orientation.…”

De novo design of drug-binding proteins with predictable binding energy and specificity

“…Another important feature is the need to consider the energetically unfavorable loss of hydrogen bonds between water molecules and both the drug as well as the protein’s binding site, which occurs when the drug binds the pocket. Although COMBS and other algorithms ( 12 ) consider the need to form hydrogen bonds to compensate for the loss of hydration, we strove to form a fuller set of compensatory ligand-protein hydrogen bonds to every buried polar atom. We sampled vdMs between the ligand and the first-shell amino acids, as well as vdMs between the first shell and a second shell of interacting residues, which also assures a favorable geometry for the residues directly contacting the ligand.…”

“…Although we have an advanced understanding of both protein design and molecular interactions, the rational design of de novo proteins that specifically bind small molecules with low nanomolar to picomolar affinity is a major challenge (1, 2) that has not been achieved in de novo proteins (3,4) without experimental screening of large libraries of variants (5)(6)(7). Even with the application of recent advances in artificial intelligence to facilitate de novo design (8)(9)(10), it has been necessary to screen thousands of independent designs to discover binders with low-micromolar to high-nanomolar dissociation constants (K d ) directly from design algorithms (3,(11)(12)(13)(14). Proteins with higher affinity are often desirable.…”

Efficient Generation of Protein Pockets with PocketGen

“…Recent advancements in protein pocket design have been facilitated by deep learning-based approaches 3,8,16,[22][23][24] . For instance, RFDiffusion 25 employs denoising diffusion probabilistic models 26 in conjunction with RoseTTAFold 27 for de novo protein structure generation.…”

“…These methods, specifically devised for antibodies, face challenges when adapting to pocket designs conditioned on target ligand molecules. Hybrid approaches that combine deep learning models with traditional methods are also being explored 3,8 . For example, Yeh et al 8 developed a novel Luciferase by employing a combination of protein hallucination 34 , the trRosetta structure prediction neural network 35 , hydrogen bonding networks, and RifDock 36 , generating a multitude of idealized protein structures with diverse pocket shapes for subsequent filtering.…”