37. Capacity of Viral Genome Packaging and Internal Volumes of AAV Viral Particles

Abstract

Adeno-associated virus (AAV) vectors have widely been used in many preclinical and clinical studies, showing promise as the most effective in vivo gene delivery vector. Despite successful application of AAV vector-mediated gene therapy to the treatment of certain genetic diseases such as hemophilia, the inability to package genomes of longer than 5 kb has precluded broader applications of this approach. Whether or not longer genomes would become able to be packaged by viral capsid engineering has yet to be determined; however, what is lacking in this field is comprehensive understanding of the maximal capacity of genome packaging in various AAV serotypes through a systematic analysis. Here we developed a novel computational algorithm to calculate the capsid internal volume that minimizes approximation and determine the internal volumes of 22 eukaryotic viruses with T=1 icosahedral symmetry including 16 viruses that belong to the family of Parvoviridae. In addition, we systematically determine the maximal packaging capacity of both VP1/VP2/VP3 capsids and VP3 only capsids derived from various serotypes using a panel of recombinant AAV2 vector genomes of varying sizes from 4.80 kb up to 6.00 kb by an increment of 0.05 kb, termed ruler genomes. To determine capsid internal volumes, three dimensional structure coordinates of each viral capsid were retrieved from the Protein Data Bank, and then, using an algorithm we have developed, each atom was inflated in such a way that interstices between adjacent atoms in the protein shell are effectively filled with protein while keeping the shape of the inner surface of the capsid unchanged. The internal volume of the capsid was then calculated by a numerical integration of coordinates in the non-protein space that were closer to the center than the atom that is most distant from the center of the capsid. By this method, we find that the genome lengths and internal volumes of the 22 eukaryotic T=1 viruses that we analyzed vary from 0.83 kb to 6.04 kb and from 6.72 × 105 to 3.46 × 106 A3, respectively, showing a strong non-linear allometric relationship with an exponent of 1.30 (R2=0.97). In the following AAV serotypes, AAV1, 2, 3B, 4, 5, 6, 8 and 9, AAV5 is found to have a distinctively smaller volume (2.48 × 106 A3) compared to the other 7 serotypes (2.62±0.03 × 106 A3, mean±SD). This analysis also reveals that AAV2 VP3 only capsid has a 7% larger volume than AAV2 VP1/VP2/VP3 capsid. In keeping with this observation, a preliminary study using the ruler genomes has indicated that the AAV2 VP3 capsid can package a genome that is longer than that can be packaged in the AAV2 VP1/VP2/VP3 capsid by several hundred nucleotides, which appears to correspond to the difference in the volume of the two different capsids. Investigation of other serotypes and capsid mutants is currently underway. Thus, the approach presented here will help understand the genome packaging and provide insights into how to design capsids to overcome packaging limitations.