Adeno-associated viruses (AAVs) are a leading vector for gene therapy, yet their clinical utility is limited by the lack of robust quality control methods to distinguish between empty (AAV
empty), partially loaded (AAV
partial), and fully DNA loaded (AAV
full) capsids. Current analytical techniques provide partial insights but remain limited in sensitivity, throughput, or resolution. Here we present a multimodal plasmonic nanopore sensor that integrates optical trapping with electrical resistive-pulse sensing to characterize AAV9 capsids at the single-particle level in tens of μL sample volumes and fM range concentrations. As a model system, we employed AAV9 capsids not loaded with DNA, capsids loaded with a self-complementary 4.7 kbp DNA (AAV
scDNA), and ones loaded with single-stranded 4.7 kbp DNA (AAV
ssDNA). Ground-truth validation was performed with analytical ultracentrifugation (AUC). Nanosensor data were acquired concurrently for optical step changes (occurring at AAV trapping and un-trapping) both in transmittance and reflectance geometries, and electrical nanopore resistive pulse signatures, making for a total of five data dimensions. The acquired data was then filtered and clustered by Gaussian mixture models (GMMs), accompanied by spectral clustering stability analysis, to successfully separate between AAV species based on their DNA load status (AAV
empty, AAV
partial, AAV
full) and DNA load type (AAV
scDNA versus AAV
ssDNA). The motivation for quantifying the AAV
empty and AAV
partial population fractions is that they reduce treatment efficacy and increase immunogenicity. Likewise, the motivation to identify AAV
scDNA population fractions is that these have much higher transfection rates. Importantly, the results showed that the nanosensor could differentiate between AAV
scDNA and AAV
ssDNA despite their identical masses. In contrast, AUC could not differentiate between AAV
scDNA and AAV
ssDNA. An equimolar mixture of AAV
scDNA, AAV
ssDNA and AAV
empty was also measured with the sensor, and the results showed the expected population fractions, supporting the capacity of the method to differentiate AAV load status in heterogeneous solutions. In addition, less common optical and electrical signal signatures were identified in the acquired data, which were attributed to debris, rapid entry re-entry to the optical trap, or weak optical trap exits, representing critical artifacts to recognize for correct interpretation of the data. Together, these findings establish plasmonic nanopore sensing as a promising platform for quantifying AAV DNA loading status and genome type with the potential to extend ultra-sensitive single-particle characterization beyond the capabilities of existing methods.
Full article