Â鶹ÒùÔº

Understanding molecular diversity is fundamental to biomedical research and diagnostics, but existing analytical tools struggle to distinguish subtle variations in the structure or composition among biomolecules, such as proteins. Researchers at the University of Tokyo have developed a new analytical approach which helps overcome this problem. The new method, called voltage-matrix nanopore profiling, combines multivoltage solid-state nanopore recordings with machine learning for accurate classification of proteins in complex mixtures, based on the proteins' intrinsic electrical signatures.

The study, in Chemical Science, demonstrates how this new framework can identify and classify "molecular individuality" without the need for labels or modifications. The research holds promise of providing a foundation that could lead to more advanced and wider applications of molecular analysis in various areas, including disease diagnosis.

Solid-state nanopores are tiny tunnels that a protein or other molecule can pass through, driven by the ionic current through the opening. By applying voltage to this process, signals produced as the molecules pass through the nanopores can help identify the molecule. While nanopore technologies have revolutionized DNA and RNA analysis, their application to proteins has been limited due to the proteins' complex structures and variability in signal behavior.

The team's new approach systematically varies voltage conditions, capturing both stable and voltage-dependent signal patterns. Organizing these features into a voltage matrix enables a machine learning model to distinguish proteins even within mixtures, extending the use of nanopore measurements beyond sequencing toward general molecular profiling.

"Identifying and classifying proteins within complex biological mixtures is difficult. Traditional methods like enzyme-linked immunosorbent assay (ELISA) or mass spectrometry often struggle to resolve subtle structural differences or dynamic states, especially without labeling," said Professor Sotaro Uemura in the Department of Biological Sciences at the University of Tokyo. "Solid-state nanopores provide a promising solution, but previous approaches were limited by their reliance on single-voltage measurements. Our work set out to overcome these limitations."

To demonstrate the concept, the researchers analyzed mixtures containing two cancer-related protein biomarkers, carcinoembryonic antigen (CEA) and cancer antigen 15-3 (CA15-3). By constructing a voltage matrix from signals recorded under six voltage conditions, they identified distinct response patterns characteristic of each protein. The approach also detected shifts in molecular populations when an aptamer, a short, synthetic DNA segment, was bound to CEA.

Furthermore, to examine the practicality of the approach, the researchers applied the voltage-matrix framework to mouse serum samples. By comparing sera that had or had not undergone centrifugation, and analyzing them under multiple voltage conditions, they found that the two types of samples could be clearly distinguished within the voltage matrix. This result indicates that the method can detect and classify subtle compositional differences in complex, biologically derived samples, supporting its potential applicability to real-world bioanalytical and diagnostic contexts.

"By systematically varying voltage conditions and applying machine learning, we can create a voltage-matrix that reveals both robust, voltage-independent molecular features and voltage-sensitive structural changes," said Uemura. "Our study is not simply about improving detection sensitivity—it establishes a new way to represent and classify molecular signals across voltages, allowing us to visualize molecular individuality and estimate compositions within mixtures."

Looking ahead, the team plans to extend the framework to human serum or saliva samples and to develop a parallelized nanopore system, carrying out multiple tasks simultaneously, for real-time molecular profiling—a foundation that could ultimately support applications from biomedical diagnostics to monitoring environmental changes.