DNA増幅によるECLおよびCVデュアルモード信号を利用したタンパク質凝集およびROS産生抑制のための低分子医薬品のスクリーニング支援における機械学習

Screening of small-molecule drugs to suppress both protein aggregation and reactive oxygen species (ROS) generation is critical for developing therapies for neurodegenerative diseases (NDs). However, existing methods are limited to characterizing only a single pathological feature (either protein aggregation or ROS generation) in a single measurement. Herein, taking α-synuclein (α-Syn) as the template protein, we developed a dual-mode electrochemical sensing platform for concurrently monitoring protein aggregation and ROS generation characteristics. A gold electrode functionalized with α-Syn via self-assembled monolayers (SAMs) was constructed as the sensing platform, realizing both ordered α-Syn immobilization and monitoring of metal ion (e.g., Cu(ii))-driven aggregation. This was accomplished by synchronously recording the electrochemiluminescence (ECL) and cyclic voltammetry (CV) dual responses of the tris(2,2'-bipyridine) ruthenium(ii) (Ru(bpy)₃ ²⁺) reporter in a single integrated assay. The catalysis of DNA oxidation by Ru(bpy)₃ ²⁺ enables the amplification of ECL and CV dual-mode signals, which increased the detection sensitivity for both aggregation and ROS generation accompanied by the α-Syn - Cu(ii) complex. Machine learning algorithms were then utilized to analyze ECL and CV responses of small molecules with known drug effects. This analysis culminated in the development of a linear discriminant analysis (LDA) screening model, which enabled the assessment of drug efficacy against the two pathological features. The predictive capability of the screening model was verified through transmission electron microscopy (TEM), cell viability and intracellular aggregation studies. This model was further successfully applied to assess two previously unexplored small molecules: diethylenetriaminepentaacetic dianhydride (DTPA) and deferiprone. Collectively, this dual-mode sensing platform, integrating DNA-amplified monitoring of protein aggregation and ROS generation, enables the robust establishment of a machine learning-assisted small-molecule drug screening model, offering a novel approach for the in vitro characterization of protein-related pathological features.