Abstract
Abstract
Acetaldehyde is a toxic pollutant that can be detoxified by acetaldehyde dehydrogenases (ADAs) through its
conversion to acetyl-CoA. This study developed an integrated approach combining virtual screening, rational design, and a dual
scoring mechanism to identify and engineer hyperactive ADA variants. A library of 5000 Dickeya parazeae ADA (DpADA)
homologues was created through protein BLAST, and deep learning tools predicted their Kcat values. The top 100 candidates were
selected based on acetaldehyde binding affinity, evaluated through molecular docking and phylogenetic analysis. Among these,
ADA6 from Buttiauxella sp. S04-F03 exhibited the highest activity, converting 57.6\% of acetaldehyde to acetyl-CoA, which was 14.1
times higher than DpADA. To improve ADA6’s thermostability, folding engineering was applied, resulting in the P443C variant with
an 80.7\% increase in residual activity after heat treatment. Molecular dynamics simulation pinpointed I440 as a bottleneck in the
substrate tunnel, guiding the design of a dual-scoring system that integrates structural adjustments and electronic optimization to
evaluate mutations for improved substrate exposure and activity. The final optimized variant, P443C-I440T, achieved a conversion
efficiency of 93.2\%. This study demonstrates the effectiveness of combining computational tools and rational mutagenesis to
enhance enzyme activity and stability in enzyme engineering.


Presenter: Hanjin Kim (M.D.)
Date: 2025.03.20 (THU) 19:00 ~ 23:00

김한진 학생이 주도하여, 딥러닝을 활용한 ADAs(adenosine deaminase) 엔지니어링 기반의 활성도 향상 연구를 주제로 저널 클럽을 진행하였습니다.