I am working with recombinant AAV and I am puzzled by the limit of their packaging capacity (around 5Kb ssDNA), that is really a huge drawback for gene delivery. I have tried to look more in deep into this problem but I couldn't find any paper that could clarify what is the reason of this limitation. Is it the size of the capsid or something else? I suppose that it is the size of the capsid but there are smaller capsids able to accommodate more DNA so it could be due to other factors. Either way, I could not find the proper literature to solve my doubts. Do you know the answer? Or, can you point me to good papers on this topic?
It's not the size that counts, it's how you pack it (sorry couldn't help myself).
Wild-type bidensoviruses do a lot of interesting things to pack genome like splitting its genome into two virions(1), inverted terminal repeats (ITRs) that form panhandles that can stack instead of hairpins (2), and overlapping positive and negative ssDNA strand ends (2 and 3) that clip closely together. So on the whole, I don't know that it's a fair comparison.
Also remember that you can go too small with AAV genome size, not just too large (if you go too small you will even get 2 genomes per virion).
If you want to orderly pack ssDNA (or DNA in general), you have to use some nucleoproteins to help. Free ssDNA has the added problem of being unstable. Going back to basic biology, you can think of nucleosomes allowing the packing of eukaryotic DNA at a much higher density than would otherwise be energetically favorable.
Viruses use a wide variety of methods to pack genetic material, one of which is through binding directly with the capsid as with AAVs. One should remember that the capsid proteins also are tasked with entry into the cell, trafficking, and entry into the nucleus. All while protecting and packaging the genetic payload.
AAVs, being small (wt is ~4.7kb w/7 proteins), only have the capsid proteins VP1-3 to work with. The ITRs are known to be the cis active sequences of the genome (4, 5), and these ITRs form normal hairpins.
A portion of the AAV ITR binds (hooks onto) an interior facing portion of the VP3 gene (6, 7). But the virus has to push in this ssDNA, and it uses the Rep proteins to do it (8). The Rep proteins (Rep52 and Rep78 specifically), are larger than the capsid proteins, and bind the ITRs at multiple sites.
While the exact insertion pore hasn't been identified at the structural level, the way I like to picture it is a series of hairpins have to bind on spokes of the Rep proteins to be pushed into the virion, where in turn the hairpins have to hook onto the inside of the capsid. If you put too much ssDNA between your ITRs, then you lose the ability to be passed along (or bind) to both proteins in packaging. You start to lose the tight packing onto VP3 found in wt virus, and instead get more sporadic VP3 binding (every 3rd as opposed to every other).
It would be incredibly hard to just "increase" the size of the capsid without messing up the assembly and the other functions of the capsid. The VPs only oligomerize in the cytoplasm (9) suggesting that host proteins are required for proper assembly. I think the process of successfully increasing the optimal genome size would be quite a task and achievement if done. One would likely have to optimize additional ITRs, the internal binding bind sites on VP3, and the structure of Rep52 before such a thing would be possible. Alternatively bringing in a novel or chimeric packing mechanism, but again this would be exceedingly difficult (probably prize worthy).