AI解析によるタンパク質設計原理の解明(Protein Design Principles Revealed in AI Analysis of Nanoribbon Assembly)

ad

2026-03-16 パシフィック・ノースウェスト国立研究所(PNNL)

パシフィック・ノースウェスト国立研究所(PNNL)の研究は、AI解析を用いてタンパク質ナノリボン構造の形成原理を解明した。機械学習により、アミノ酸配列と自己組織化構造との関係を解析し、特定の配列パターンが安定したナノリボン形成を導くことを明らかにした。これにより、従来は経験的だったタンパク質設計に理論的指針が与えられ、ナノ材料やバイオ材料の精密設計が可能となる。成果は、医療材料やバイオナノテクノロジー分野への応用が期待される。

<関連情報>

溶媒力と対称性の破れが、液固界面における設計タンパク質の集合に及ぼす影響 Impact of solvent forces and broken symmetry on the assembly of designed proteins at a liquid-solid interface

Sakshi Yadav Schmid,Benjamin Helfrecht,Amy Stegmann,Benjamin A. Legg,Harley Pyles,Jiajun Chen,John R. Edison,Maxim Ziatdinov,Zdenek Preisler,Orion Dollar,Stephen Whitelam,Sergei Kalinin,David Baker,Christopher J. Mundy,Shuai Zhang & James J. De Yoreo
Nature Communications  Published:13 March 2026
DOI:https://doi.org/10.1038/s41467-026-69170-0

AI解析によるタンパク質設計原理の解明(Protein Design Principles Revealed in AI Analysis of Nanoribbon Assembly)

Abstract

The era of protein design has enabled the creation of hybrid protein-inorganic interfaces, leading to both surface-directed self-assembly of de novo protein architectures and protein-directed formation of inorganic materials. However, the resulting patterns of protein assembly are often unexpected, implying that essential interactions are not accounted for in current design platforms. Here, we use high-speed atomic force microscopy (AFM) analyzed through machine learning to follow the assembly of protein nanorods in aqueous electrolytes on two types of mica exhibiting disparate symmetry elements, which are imprinted on the overlying hydration structure. Using Monte Carlo simulations, we reproduce the observed phases and show that an observed smectic phase, previously thought to be unstable for non-interacting rods in two dimensions, emerges when crystal symmetry introduces a directional bias. The findings demonstrate the importance of incorporating solvent forces as modulated by the hydration structure inherent to interfacial systems when designing protein assemblies at liquid-crystal interfaces. Coupling physics-based simulations that can account for these factors to de novo protein design algorithms can lead to improved design platforms for bio-inspired, hybrid materials.

生物化学工学
ad
ad
Follow
ad
タイトルとURLをコピーしました