Some suggest measuring refractive index, some suggest collecting all fractions and selecting virus-rich fraction by qPCR.
I'm wondering why after purification I do not see the formation of a distinct band, like adenovirus, after the same procedure?.
AAV does form a band after CsCl. You should actually see two bands, one for the empty capsid, and one for full packaged capsid. During routine AAV purifications, the virus containing bands are pulled and then subjected to a second isopycnic CsCl spin to allow for further separation of the empty and full bands.
Thank you so much, John!
May I ask what the CsCl concentration you use?
For CsCl gradient spin, we don't see any band. What we did was to collect each 4ml fraction and determine which fraction to use by RT-PCR. It was labor-intense.
We are now using iodixanol step gradient. We can see a transparent band in 40% if look carefully.
We use a step gradient with 5 mls of 1.5 g/mL on the bottom, 12 mls of 1.3 g/mL in the middle, and then layer our vector containing sample (15-20 mls) on top. After the overnight spin we pull the vector bands (can be hard to separate the empty and full bands), add them to a new tube, fill the tube with 1.4 g/mL CsCl and then do an isopycnic spin overnight. This should give two nice bands. Note that we do this routinely for AAV5, but this will also be the same for AAV8.
If your titers are low, then it will be difficult to see a band. We use a bright light under the tube to help us visualize the band. You will need to have a titer of at least 1e13 vg to see a reasonable band, and the more the better.
From my experience it is hard/impossible (?) to see a distinct band after running a continuous CsCl gradient. We indeed routinely go for refractive index measurement which works well and is pretty fast, too. In general I would recommend to do an initial gradient analysis measuring AAV capsids (an AAV8 capsid ELISA is available meanwhile), viral genomes (qPCR) and RI to establish your system. As soon as you know for your set-up which RI corresponds to which fraction, it should be a quite reproducible way of harvesting the right fractions.
Apart from that, you might want to think about using a Iodixanol-based step gradient. The protocol is much faster (only 2 h of ultracentrifugation) and in our hands works equally well with regard to purity and vector yield.
We usually only have trouble seeing a band on a continuous gradient when there is a mixed population of genomes. For instance, if you are trying to package a small genome, you will generally get a mixed population containing fragments of various genome sizes. These will all have different densities and thus will run as a smear. This generally isn't a big problem when you are using a full sized (~5kb genome). Iodixanol is also good, but does not give good separation between empties and fulls (at least not in our hands).
For my second trial, I did see a distinct and wide band after the first CsCl gradient, which unfortunately become a smear on the second CsCl gradient. It may be due to the inconsistent concentration for the two runs???
John, I don't understand why you get a mixed population when packaging a small genome. Our GOI is 2.3 kb, do you consider it a small genome to package?
Ben, for the RI, do you take each fraction out and measure the RI, and decide which one is in the right range. I assume you need to dialyze afterwards to get rid of the salt?
If you have a small genome (2.3 kb), you can be sure that you will have a mixed population containing some that has 2.3 kb, and some that has 4.6 kb, and usually a mix of sizes in between. Is your vector double stranded or single stranded? If is ds, you might only have one band around 4.6 kb. Have you run your vector out on an alkaline agarose gel? That will give you an idea of the genome sizes that you have. AAV likes to have full capsid and will package pretty much as much as it can. There is a good reference (Dong et al, Human Gene therapy 1996) on this. Most people never run their vector genomes out on an agarose gel so they never realize this.
Thanks for the clarification. My vector should be ssDNA virus since it's not the scAAV,
I wanna make sure I understand it correctly, when we talk about genome size, it should be LTR to LTR, right? In my case, LTR+promoter+goi+other element+LTR, total is more than 4 kb, which pretty much reachesthe packaging limit of AAV, do you think we still could get a mixed population?
Yes, we measure the RI of each fraction. However, in practice you will know very soon which fractions will probably be in the right RI range, so you will actually only have to measure 4 or 5 fractions to identify the desired upper and lower limit of the RI range.
Regarding dialyis, we carry out at least 3 rounds against 1xPBS. We then finally concentrate (ultrafiltration) and sterile filtrate the AAVs.
Yes, genome size is ITR to ITR (FYI, LTRs are the ends of lentiviral vectors, not AAV). 4 kb should give a fairly uniform population. The best way to know is to run 5 uL of your vector on an alkaline gel and see.
- Badges
- Science method