Collaborative Research: Multimodal Nanopore Biosensing and Nanosieving for Upstream Quality Control of AAV Gene Therapy Vectors

Gene therapy holds the promise of curing genetic diseases by delivering corrective DNA into patients’ cells using tiny protein shells called adeno-associated viruses (AAVs). For safe and effective treatments, the viral shells must be properly loaded with the therapeutic DNA. However, during manufacturing, many shells come out empty or only partially filled, and there is no way to check DNA loading quality until after production is complete. Poor-quality batches are discovered late, wasting millions of dollars and potentially putting patients at risk. This project will develop a new type of sensor that can detect, at the level of a single virus particle, whether each capsid is fully loaded, partially loaded, or empty. The sensor will combine a focused laser beam with tiny electrical channels to trap and measure individual viral particles. A computer program will then analyze the combined light and electrical signals to classify each particle. To enable the sensor in complex solutions, the team will develop miniaturized filtration tools to remove debris and concentrate the particles of interest before measurement. This work will advances national health by making gene therapy safer and more affordable. It will help accelerate the path to regulatory approval for life-saving treatments for rare genetic diseases. The project will also train the next generation of scientists and engineers through hands-on research experiences for undergraduate and graduate students, outreach to K-12 students and teachers in the Dallas-Fort Worth area, and the development of an interactive haptic simulator that lets students physically experience the forces acting on nanoscale biological particles. This project will develop and validate an integrated optical-electrical nanopore biosensing platform for single-particle quantification of adeno-associated virus (AAV) DNA loading efficiency at upstream stages of biomanufacturing, a capability that does not currently exist. The project will determine whether concurrent optical and electrical nanopore sensing can resolve fully loaded, partially loaded, and genome-empty capsids in complex pre-purification mixtures, enabling predictive quality control (QC) before scale-up. The platform will integrate three synergistic capabilities. First, a self-induced back-action actuated nanopore electrophoresis sensor, comprising a double nanohole plasmonic structure positioned above a solid-state nanopore, will provide simultaneous optical detection of particle size and refractive index, direct-current nanopore sensing of capsid charge and deformability, and alternating-current impedance spectroscopy. Second, to enable analysis in complex upstream matrices, microfluidic isotachophoresis will separate AAV particles from small-molecule debris, and solid-state nanopore nanosieves will selectively deplete AAVempty, reducing its 20-fold excess relative to AAVfull in clarified lysates. Third, a machine learning (ML) framework employing a one-dimensional U-net convolutional neural network for feature extraction followed by a Bayesian deep neural network for calibrated classification will transform the high-dimensional, multimodal signal datasets into quantitative, particle-level predictions with associated uncertainty estimates. Classifiers will be evaluated using per-class precision, recall, F1 score, area under the precision-recall curve, and calibration error, across sample compositions reflecting the class imbalance typical of upstream manufacturing stages. Performance will be validated against AUC, transmission electron microscopy, atomic force microscopy, and quantitative polymerase chain reaction using benchmark samples provided by the UT Southwestern Medical Center cGMP AAV Core Facility. Successful completion will establish the first all-in-one nanosensing and nanosieving platform for upstream AAV DNA load QC, with broad applicability to other viral vectors, exosomes, and lipid nanoparticles. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.